enabling cognitive workloads on the cloud: gpus with mesos, docker and marathon on power
TRANSCRIPT
ibmedge copy 2016 IBM Corporation
Enabling Cognitive Workloads on the Cloud GPUs with Mesos Docker and Marathon on POWER
Seetharami Seelam IBM Research
Indrajit Poddar IBM Systems
ibmedge
Please Note
bull IBMrsquos statements regarding its plans directions and intent are subject to change or withdrawal without notice and at IBMrsquos sole discretion
bull Information regarding potential future products is intended to outline our general product direction and it should not be relied on in making a purchasing decision
bull The information mentioned regarding potential future products is not a commitment promise or legal obligation to deliver any material code or functionality Information about potential future products may not be incorporated into any contract
bull The development release and timing of any future features or functionality described for our products remains at our sole discretion
bull Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment The actual throughput or performance that any user will experience will vary depending upon many factors including considerations such as the amount of multiprogramming in the userrsquos job stream the IO configuration the storage configuration and the workload processed Therefore no assurance can be given that an individual user will achieve results similar to those stated here
1
ibmedge
About Seelam
Expertise bull 10+ years in Large scale high performance and
distributed systems
bull Built multiple cloud services for IBM Bluemix autoscaling business rules containers POWER containers and Deep Learning as a service
bull Enabled and scaled Docker on POWERZ for extreme density (tens of thousands of containers)
bull Enabling GPUs in the cloud for container-based workloads (MesosKubDocker)
2
Dr Seetharami R Seelam Research Staff Member IBM T J Watson Research Center Yorktown Heights NY sseelamusibmcom Twitter sseelam
ibmedge
About Indrajit (aka IP)
Expertise
bull Accelerated Cloud Data Services Machine
Learning and Deep Learning bull Apache Spark TensorFlowhellip with GPUs
bull Distributed Computing (scale out and up)
bull Cloud Foundry Spectrum Conductor Mesos Kubernetes Docker OpenStack WebSphere
bull Cloud computing on High Performance Systems
bull OpenPOWER IBM POWER
3
Indrajit Poddar Senior Technical Staff Member Master Inventor IBM Systems ipoddarusibmcom Twitter ipoddar
ibmedge
Agenda
bull Introduction to Cognitive workloads and POWER
bull Requirements for GPUs in the Cloud
bull MesosGPU enablement
bull KubernetesGPU enablement
bull Demo of GPU usage with Docker on OpenPOWER to identify dog breads
bull Machine and Deep Leaning Ecosystem on OpenPOWER
bull Summary and Next Steps
4
ibmedge
Cognition
5
What you and I (our brains) do without even thinking about ithellipwe recognize a bicycle
ibmedge
Now machines are learning the way we learnhellip
6
From Texture of the Nervous System
of Man and the Vertebrates by
Santiago Ramoacuten y Cajal Artificial Neural Networks
ibmedge
But training needs a lot computational resources
Easy scale-out with Deep Learning model training is hard to distribute
Training can take hours days or weeks
Input data and model sizes are becoming
larger than ever (eg video input billions of
features etc)
Real-time analytics with Unprecedented demand for offloaded computation
accelerators and higher memory bandwidth systems
Resulting inhellip
Moorersquos law is dying
ibmedge
OpenPOWER Open Hardware for High Performance
8
Systems designed for
big data analytics
and superior cloud economics
Upto
12 cores per cpu
96 hardware threads per cpu
1 TB RAM
76Tbs combined IO Bandwidth
GPUs and FPGAs cominghellip
OpenPOWER
Traditional
Intel x86
httpwwwsoftlayercombare-metal-searchprocessorModel[]=9
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Please Note
bull IBMrsquos statements regarding its plans directions and intent are subject to change or withdrawal without notice and at IBMrsquos sole discretion
bull Information regarding potential future products is intended to outline our general product direction and it should not be relied on in making a purchasing decision
bull The information mentioned regarding potential future products is not a commitment promise or legal obligation to deliver any material code or functionality Information about potential future products may not be incorporated into any contract
bull The development release and timing of any future features or functionality described for our products remains at our sole discretion
bull Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment The actual throughput or performance that any user will experience will vary depending upon many factors including considerations such as the amount of multiprogramming in the userrsquos job stream the IO configuration the storage configuration and the workload processed Therefore no assurance can be given that an individual user will achieve results similar to those stated here
1
ibmedge
About Seelam
Expertise bull 10+ years in Large scale high performance and
distributed systems
bull Built multiple cloud services for IBM Bluemix autoscaling business rules containers POWER containers and Deep Learning as a service
bull Enabled and scaled Docker on POWERZ for extreme density (tens of thousands of containers)
bull Enabling GPUs in the cloud for container-based workloads (MesosKubDocker)
2
Dr Seetharami R Seelam Research Staff Member IBM T J Watson Research Center Yorktown Heights NY sseelamusibmcom Twitter sseelam
ibmedge
About Indrajit (aka IP)
Expertise
bull Accelerated Cloud Data Services Machine
Learning and Deep Learning bull Apache Spark TensorFlowhellip with GPUs
bull Distributed Computing (scale out and up)
bull Cloud Foundry Spectrum Conductor Mesos Kubernetes Docker OpenStack WebSphere
bull Cloud computing on High Performance Systems
bull OpenPOWER IBM POWER
3
Indrajit Poddar Senior Technical Staff Member Master Inventor IBM Systems ipoddarusibmcom Twitter ipoddar
ibmedge
Agenda
bull Introduction to Cognitive workloads and POWER
bull Requirements for GPUs in the Cloud
bull MesosGPU enablement
bull KubernetesGPU enablement
bull Demo of GPU usage with Docker on OpenPOWER to identify dog breads
bull Machine and Deep Leaning Ecosystem on OpenPOWER
bull Summary and Next Steps
4
ibmedge
Cognition
5
What you and I (our brains) do without even thinking about ithellipwe recognize a bicycle
ibmedge
Now machines are learning the way we learnhellip
6
From Texture of the Nervous System
of Man and the Vertebrates by
Santiago Ramoacuten y Cajal Artificial Neural Networks
ibmedge
But training needs a lot computational resources
Easy scale-out with Deep Learning model training is hard to distribute
Training can take hours days or weeks
Input data and model sizes are becoming
larger than ever (eg video input billions of
features etc)
Real-time analytics with Unprecedented demand for offloaded computation
accelerators and higher memory bandwidth systems
Resulting inhellip
Moorersquos law is dying
ibmedge
OpenPOWER Open Hardware for High Performance
8
Systems designed for
big data analytics
and superior cloud economics
Upto
12 cores per cpu
96 hardware threads per cpu
1 TB RAM
76Tbs combined IO Bandwidth
GPUs and FPGAs cominghellip
OpenPOWER
Traditional
Intel x86
httpwwwsoftlayercombare-metal-searchprocessorModel[]=9
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
About Seelam
Expertise bull 10+ years in Large scale high performance and
distributed systems
bull Built multiple cloud services for IBM Bluemix autoscaling business rules containers POWER containers and Deep Learning as a service
bull Enabled and scaled Docker on POWERZ for extreme density (tens of thousands of containers)
bull Enabling GPUs in the cloud for container-based workloads (MesosKubDocker)
2
Dr Seetharami R Seelam Research Staff Member IBM T J Watson Research Center Yorktown Heights NY sseelamusibmcom Twitter sseelam
ibmedge
About Indrajit (aka IP)
Expertise
bull Accelerated Cloud Data Services Machine
Learning and Deep Learning bull Apache Spark TensorFlowhellip with GPUs
bull Distributed Computing (scale out and up)
bull Cloud Foundry Spectrum Conductor Mesos Kubernetes Docker OpenStack WebSphere
bull Cloud computing on High Performance Systems
bull OpenPOWER IBM POWER
3
Indrajit Poddar Senior Technical Staff Member Master Inventor IBM Systems ipoddarusibmcom Twitter ipoddar
ibmedge
Agenda
bull Introduction to Cognitive workloads and POWER
bull Requirements for GPUs in the Cloud
bull MesosGPU enablement
bull KubernetesGPU enablement
bull Demo of GPU usage with Docker on OpenPOWER to identify dog breads
bull Machine and Deep Leaning Ecosystem on OpenPOWER
bull Summary and Next Steps
4
ibmedge
Cognition
5
What you and I (our brains) do without even thinking about ithellipwe recognize a bicycle
ibmedge
Now machines are learning the way we learnhellip
6
From Texture of the Nervous System
of Man and the Vertebrates by
Santiago Ramoacuten y Cajal Artificial Neural Networks
ibmedge
But training needs a lot computational resources
Easy scale-out with Deep Learning model training is hard to distribute
Training can take hours days or weeks
Input data and model sizes are becoming
larger than ever (eg video input billions of
features etc)
Real-time analytics with Unprecedented demand for offloaded computation
accelerators and higher memory bandwidth systems
Resulting inhellip
Moorersquos law is dying
ibmedge
OpenPOWER Open Hardware for High Performance
8
Systems designed for
big data analytics
and superior cloud economics
Upto
12 cores per cpu
96 hardware threads per cpu
1 TB RAM
76Tbs combined IO Bandwidth
GPUs and FPGAs cominghellip
OpenPOWER
Traditional
Intel x86
httpwwwsoftlayercombare-metal-searchprocessorModel[]=9
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
About Indrajit (aka IP)
Expertise
bull Accelerated Cloud Data Services Machine
Learning and Deep Learning bull Apache Spark TensorFlowhellip with GPUs
bull Distributed Computing (scale out and up)
bull Cloud Foundry Spectrum Conductor Mesos Kubernetes Docker OpenStack WebSphere
bull Cloud computing on High Performance Systems
bull OpenPOWER IBM POWER
3
Indrajit Poddar Senior Technical Staff Member Master Inventor IBM Systems ipoddarusibmcom Twitter ipoddar
ibmedge
Agenda
bull Introduction to Cognitive workloads and POWER
bull Requirements for GPUs in the Cloud
bull MesosGPU enablement
bull KubernetesGPU enablement
bull Demo of GPU usage with Docker on OpenPOWER to identify dog breads
bull Machine and Deep Leaning Ecosystem on OpenPOWER
bull Summary and Next Steps
4
ibmedge
Cognition
5
What you and I (our brains) do without even thinking about ithellipwe recognize a bicycle
ibmedge
Now machines are learning the way we learnhellip
6
From Texture of the Nervous System
of Man and the Vertebrates by
Santiago Ramoacuten y Cajal Artificial Neural Networks
ibmedge
But training needs a lot computational resources
Easy scale-out with Deep Learning model training is hard to distribute
Training can take hours days or weeks
Input data and model sizes are becoming
larger than ever (eg video input billions of
features etc)
Real-time analytics with Unprecedented demand for offloaded computation
accelerators and higher memory bandwidth systems
Resulting inhellip
Moorersquos law is dying
ibmedge
OpenPOWER Open Hardware for High Performance
8
Systems designed for
big data analytics
and superior cloud economics
Upto
12 cores per cpu
96 hardware threads per cpu
1 TB RAM
76Tbs combined IO Bandwidth
GPUs and FPGAs cominghellip
OpenPOWER
Traditional
Intel x86
httpwwwsoftlayercombare-metal-searchprocessorModel[]=9
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Agenda
bull Introduction to Cognitive workloads and POWER
bull Requirements for GPUs in the Cloud
bull MesosGPU enablement
bull KubernetesGPU enablement
bull Demo of GPU usage with Docker on OpenPOWER to identify dog breads
bull Machine and Deep Leaning Ecosystem on OpenPOWER
bull Summary and Next Steps
4
ibmedge
Cognition
5
What you and I (our brains) do without even thinking about ithellipwe recognize a bicycle
ibmedge
Now machines are learning the way we learnhellip
6
From Texture of the Nervous System
of Man and the Vertebrates by
Santiago Ramoacuten y Cajal Artificial Neural Networks
ibmedge
But training needs a lot computational resources
Easy scale-out with Deep Learning model training is hard to distribute
Training can take hours days or weeks
Input data and model sizes are becoming
larger than ever (eg video input billions of
features etc)
Real-time analytics with Unprecedented demand for offloaded computation
accelerators and higher memory bandwidth systems
Resulting inhellip
Moorersquos law is dying
ibmedge
OpenPOWER Open Hardware for High Performance
8
Systems designed for
big data analytics
and superior cloud economics
Upto
12 cores per cpu
96 hardware threads per cpu
1 TB RAM
76Tbs combined IO Bandwidth
GPUs and FPGAs cominghellip
OpenPOWER
Traditional
Intel x86
httpwwwsoftlayercombare-metal-searchprocessorModel[]=9
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Cognition
5
What you and I (our brains) do without even thinking about ithellipwe recognize a bicycle
ibmedge
Now machines are learning the way we learnhellip
6
From Texture of the Nervous System
of Man and the Vertebrates by
Santiago Ramoacuten y Cajal Artificial Neural Networks
ibmedge
But training needs a lot computational resources
Easy scale-out with Deep Learning model training is hard to distribute
Training can take hours days or weeks
Input data and model sizes are becoming
larger than ever (eg video input billions of
features etc)
Real-time analytics with Unprecedented demand for offloaded computation
accelerators and higher memory bandwidth systems
Resulting inhellip
Moorersquos law is dying
ibmedge
OpenPOWER Open Hardware for High Performance
8
Systems designed for
big data analytics
and superior cloud economics
Upto
12 cores per cpu
96 hardware threads per cpu
1 TB RAM
76Tbs combined IO Bandwidth
GPUs and FPGAs cominghellip
OpenPOWER
Traditional
Intel x86
httpwwwsoftlayercombare-metal-searchprocessorModel[]=9
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Now machines are learning the way we learnhellip
6
From Texture of the Nervous System
of Man and the Vertebrates by
Santiago Ramoacuten y Cajal Artificial Neural Networks
ibmedge
But training needs a lot computational resources
Easy scale-out with Deep Learning model training is hard to distribute
Training can take hours days or weeks
Input data and model sizes are becoming
larger than ever (eg video input billions of
features etc)
Real-time analytics with Unprecedented demand for offloaded computation
accelerators and higher memory bandwidth systems
Resulting inhellip
Moorersquos law is dying
ibmedge
OpenPOWER Open Hardware for High Performance
8
Systems designed for
big data analytics
and superior cloud economics
Upto
12 cores per cpu
96 hardware threads per cpu
1 TB RAM
76Tbs combined IO Bandwidth
GPUs and FPGAs cominghellip
OpenPOWER
Traditional
Intel x86
httpwwwsoftlayercombare-metal-searchprocessorModel[]=9
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
But training needs a lot computational resources
Easy scale-out with Deep Learning model training is hard to distribute
Training can take hours days or weeks
Input data and model sizes are becoming
larger than ever (eg video input billions of
features etc)
Real-time analytics with Unprecedented demand for offloaded computation
accelerators and higher memory bandwidth systems
Resulting inhellip
Moorersquos law is dying
ibmedge
OpenPOWER Open Hardware for High Performance
8
Systems designed for
big data analytics
and superior cloud economics
Upto
12 cores per cpu
96 hardware threads per cpu
1 TB RAM
76Tbs combined IO Bandwidth
GPUs and FPGAs cominghellip
OpenPOWER
Traditional
Intel x86
httpwwwsoftlayercombare-metal-searchprocessorModel[]=9
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
OpenPOWER Open Hardware for High Performance
8
Systems designed for
big data analytics
and superior cloud economics
Upto
12 cores per cpu
96 hardware threads per cpu
1 TB RAM
76Tbs combined IO Bandwidth
GPUs and FPGAs cominghellip
OpenPOWER
Traditional
Intel x86
httpwwwsoftlayercombare-metal-searchprocessorModel[]=9
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
New OpenPOWER Systems with NVLink
9
S822LC-hpc ldquoMinskyrdquo
2 POWER8 CPUs with 4 NVIDIAreg Teslareg P100
GPUs GPUs hooked directly to CPUs using
Nvidiarsquos NVLink high-speed interconnect httpwww-03ibmcomsystemspowerhardwares822lc-hpcindexhtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Transparent acceleration for Deep Learning on OpenPOWER and GPUs
10
Huge speed-ups
with GPUs and
OpenPOWER
httpopenpowerdevpostcom
Impressive acceleration examples bull artNet Genre classifier
bull Distributed Tensorflow for cancer detection
bull Scale up and out genetics bioinformatics
bull Full red blood cell modeling
bull Accelerated ultrasound imaging
bull Emergency service prediction
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Enabling AcceleratorsGPUs in the cloud stack
Deep Learning apps
11
Containers and images
OR
Accelerators
Clustering frameworks
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Requirements for GPUs in the Cloud
12
FunctionFeature Comments
GPUs exposed to Dockerized
applications
Apps need access to devnvidia to use the GPUs
Support for NVIDIA GPUs Both IBM Cloud and POWER systems support NVIDIA GPUs
Support Multiple GPUs per node IBM Cloud machines have up to 2 K80s (4 GPUs) and POWER nodes
have many more
Containers require no GPU drivers GPU drivers are huge and drivers in a container creates a portability
problems so we need to support to mounting GPU drivers into the
container from the host (volume injection)
GPU Isolation GPUs allocated to a workloads should be invisible to other workloads
GPU Auto-discovery Worker node agent automatically discovers the GPU types and numbers
and report to the scheduler
GPU Usage metrics GPU utilization is critical for developers so need to expose these metrics
Support for heterogeneous GPUs in a
cluster (including app to pick a GPU
type)
IBM Cloud has M60 K80 etc and different workloads need different
GPUs
GPU sharing GPUs should be isolated between workloads
GPUs should be sharable in some cases between workloads
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
NVIDIA Docker
13
Credit httpsgithubcomNVIDIAnvidia-docker
bull A docker wrapper and tools to package and GPU based apps
bull Uses drivers on the host
bull Manual GPU assignment
bull Good for single node
bull Available on POWER
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Mesos and Ecosystem
bull Open-source cluster manager
bull Enables siloed applications to be consolidated on a shared pool of resources delivering
bull Rich framework ecosystem
bull Emerging GPU support
14
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Mesos GPU support (Joint work between Mesosphere NVIDIA and IBM)
Credit Kevin Klaues Mesosphere
Mesos support for GPUs v 11 bull Mesos will support GPU in two different
frameworks ndash Unified containerizer
bull No docker support initially
bull Remove Docker daemon from the node
ndash Docker containerizer
bull Traditional executor for Docker
bull Docker container based deployment
bull On going work ndash Code to allocate GPUs at the node in docker
containerizer
ndash Code to share the state with unified containerizer
ndash Logic for node recovery (nvidia driving this work)
bull Limitations ndash No GPU sharing between docker containers
ndash Limited GPU usage information exposed in the UI
ndash Slave recovery code will evolve over time
ndash NVIDIA GPUs only
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Implementation
bull GPU shared by mesos containerizer and docker containerizer
bull mesos-docker-executor extended to handle devices isolationexposition through docker daemon
bull Native docker implementation for CPUmemoryGPUGPU driver volume management
16
Nvidia GPU
Allocator
Nvidia Volume
Manager
Mesos
Containerizer
Docker
Containerizer Docker Daemon
CPU Memory GPU GPU driver volume
mesos-docker-executor
Nvidia GPU Isolator Mesos Agent
Docker image label check
comnvidiavolumesneeded=nvidia_driver
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Mesos GPU monitor and Marathon on OpenPOWER
17
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Usage and Progress
bull Usage
bull Compile Mesos with flag configure --with-nvml=nvml-header-path ampamp make ndashj install
bull Build GPU images following nvidia-docker (httpsgithubcomNVIDIAnvidia-docker)
bull Run a docker task with additional such resource ldquogpus=1rdquo
bull Release
bull Target release 11
bull GPU allocator for docker containerizer (code review)
bull GPU isolationexposition support for msos-docker-executor (code review)
bull GPU driver volume injection (under development)
18
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Eco-system Activities
bull Marathon
bull GPU support for Mesos Containerizer in release 13
bull GPU support for Docker Containerizer ready for release (waiting for Mesos side code merge)
19
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Kubernetes
bull Open source orchestration system for Docker containers
bull Handles scheduling onto nodes in a compute cluster
bull Actively manages workloads to ensure that their state matches the users declared intentions
bull Emerging support for GPUs
20
Kubernetes
master
Docker
Engine
Docker
Engine
Docker
Engine
Host Host Host
Kubelet
Proxy
Kubelet
Proxy
Kubelet Proxy
Etcd
cluster
-API server -Scheduler -Controller Mgr
Support HA mode
Cluster state
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Kubernetes GPU support bull Design Doc for GPU support in K8s has been out for a while
ndash httpsgithubcomkuberneteskubernetesblobmasterdocsproposalsgpu-supportmd
FunctionFeature Kub Community Our Contribution
GPUs exposed to
Dockerized applications
Yes
Support for NVIDIA GPUs Yes
Support Multiple GPUs per
node
Yes a PR is
pending
Containers require no GPU
drivers
No PoC complete
GPU Isolation Yes
GPU Auto-discovery No future
GPU Usage metrics No future
Support for heterogeneous
GPUs in a cluster
(including app to pick a
GPU type)
No future
GPU sharing No future
GPU on Kubernetes updates in community httpsgithubcomkuberneteskubernetespull28216
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Status of GPUs in Mesos and Kubernetes
22
FunctionFeature NVIDIA Docker Mesos Kubernetes
GPUs exposed to Dockerized applications
Support for NVIDIA GPUs
Support Multiple GPUs per node
Containers require no GPU drivers Future
GPU Isolation
GPU Auto-discovery Future Future
GPU Usage metrics Future Future
Support for heterogeneous GPUs in a cluster (including app to pick a
GPU type)
Future Future
GPU sharing
(not controlled)
Future Future
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
copy 2016 IBM Corporation ibmedge
Demo
23
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Machine Learning and Deep Learning analytics on OpenPOWER No code changes needed
24
ATLAS
Automatically Tuned Linear Algebra
Software)
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Learn More and Get Startedhellip
25
Power-Efficient Machine Learning on
POWER Systems using FPGA Acceleration
Machine and Deep Learning on Power Systems
Register for a SuperVessel Account and take deep learning
notebooks running in docker containers a spin
httpsny1ptopenlabcombigdata_cluster
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Summary and Next Steps bull Cognitive Machine and Deep Learning workloads are everywhere
bull OpenPOWER and Accelerators will help speed up these workloads
bull Containers can be leveraged with accelerators for agile deployment of these new workloads
bull Docker Mesos and Kubernetes are making rapid progress to support accelerators
bull OpenPOWER and this emerging cloud stack makes it the preferred platform for Cognitive workloads
|
26
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Special Thanks to Collaborators
bull Kevin Klues Mesosphere
bull Yu Bo Li IBM
bull Rajat Phull NVIdia
bull Guangya Liu IBM
bull Qian Zhang IBM
bull Benjamin Mahler Mesosphere
bull Vikrama Ditya Nvidia
bull Yong Feng IBM
bull Christy L Norman Perez IBM
bull Kubernetes Team
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
copy 2016 IBM Corporation ibmedge
Thank You
Seelam ndash sseelamusibmcom
IP - ipoddarusibmcom
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
copy 2016 IBM Corporation ibmedge
Backup
29
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Notices and Disclaimers
30
Copyright copy 2016 by International Business Machines Corporation (IBM) No part of this document may be reproduced or transmitted in any form without written permission from IBM
US Government Users Restricted Rights - Use duplication or disclosure restricted by GSA ADP Schedule Contract with IBM
Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors IBM shall have no responsibility to update this information THIS DOCUMENT IS DISTRIBUTED AS IS WITHOUT ANY WARRANTY EITHER EXPRESS OR IMPLIED IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION INCLUDING BUT NOT LIMITED TO LOSS OF DATA BUSINESS INTERRUPTION LOSS OF PROFIT OR LOSS OF OPPORTUNITY IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided
IBM products are manufactured from new parts or new and used parts In some cases a product may not be new and may have been previously installed Regardless our warranty terms applyrdquo
Any statements regarding IBMs future direction intent or product plans are subject to change or withdrawal without notice
Performance data contained herein was generally obtained in a controlled isolated environments Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved Actual performance cost savings or other results in other operating environments may vary
References in this document to IBM products programs or services does not imply that IBM intends to make such products programs or services available in all countries in which IBM operates or does business
Workshops sessions and associated materials may have been prepared by independent session speakers and do not necessarily reflect the views of IBM All materials and discussions are provided for informational purposes only and are neither intended to nor shall constitute legal or other guidance or advice to any individual participant or their specific situation
It is the customerrsquos responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customerrsquos business and any actions the customer may need to take to comply with such laws IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml
ibmedge
Notices and Disclaimers Conrsquot
31
Information concerning non-IBM products was obtained from the suppliers of those products their published announcements or other publicly available sources IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance compatibility or any other claims related to non-IBM products Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products IBM does not warrant the quality of any third-party products or the ability of any such third-party products to interoperate with IBMrsquos products IBM EXPRESSLY DISCLAIMS ALL WARRANTIES EXPRESSED OR IMPLIED INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
The provision of the information contained herein is not intended to and does not grant any right or license under any IBM patents copyrights trademarks or other intellectual property right
IBM the IBM logo ibmcom Asperareg Bluemix Blueworks Live CICS Clearcase Cognosreg DOORSreg Emptorisreg Enterprise Document Management Systemtrade FASPreg FileNetreg Global Business Services reg Global Technology Services reg IBM ExperienceOnetrade IBM SmartCloudreg IBM Social Businessreg Information on Demand ILOG Maximoreg MQIntegratorreg MQSeriesreg Netcoolreg OMEGAMON OpenPower PureAnalyticstrade PureApplicationreg pureClustertrade PureCoveragereg PureDatareg PureExperiencereg PureFlexreg pureQueryreg pureScalereg PureSystemsreg QRadarreg Rationalreg Rhapsodyreg Smarter Commercereg SoDA SPSS Sterling Commercereg StoredIQ Tealeafreg Tivolireg Trusteerreg Unicareg urbancodereg Watson WebSpherereg Worklightreg X-Forcereg and System zreg ZOS are trademarks of International Business Machines Corporation registered in many jurisdictions worldwide Other product and service names might be trademarks of IBM or other companies A current list of IBM trademarks is available on the Web at Copyright and trademark information at wwwibmcomlegalcopytradeshtml