data center design / microservices

DataCenterDesignMicroservices

BobWise

SamsungSDSResearchAmerica

Copyright © 2015 Samsung SDS Co., Ltd. All rights reserved

2


This presentation is intended to provide information concerning Samsung’s efforts around containers and container orchestration. We do our best to make sure that information presented is accurate and fully up-to-date. However, the presentation may be subject to technical inaccuracies, information that is not up-to-date or typographical errors. As a consequence, Samsung does not in any way guarantee the accuracy or completeness of information provided on this presentation. Samsung reserves the right to make improvements, corrections and/or changes to this presentation at any time.

The information in this presentation or accompanying oral statements may include forward-looking statements. These forward-looking statements include all matters that are not historical facts, statements regarding the Samsung Data System' intentions, beliefs or current expectations concerning, among other things, market prospects, growth, strategies, and the industry in which Samsung operates. By their nature, forward-looking statements involve risks and uncertainties, because they relate to events and depend on circumstances that may or may not occur in the future. Samsung cautions you that forward looking statements are not guarantees of future performance and that the actual developments of Samsung, the market, or industry in which Samsung operates may differ materially from those made or suggested by the forward-looking statements contained in this presentation or in the accompanying oral statements. In addition, even if the information contained herein or the oral statements are shown to be accurate, those developments may not be indicative developments in future periods.

Logos remain the property of their respective owners. So there.

3


This presentation is intended to provide information concerning Samsung’s efforts around containers and container orchestration. We do our best to make sure that information presented is accurate and fully up-to-date. However, the presentation may be subject to technical inaccuracies, information that is not up-to-date or typographical errors. As a consequence, Samsung does not in any way guarantee the accuracy or completeness of information provided on this presentation. Samsung reserves the right to make improvements, corrections and/or changes to this presentation at any time.

The information in this presentation or accompanying oral statements may include forward-looking statements. These forward-looking statements include all matters that are not historical facts, statements regarding the Samsung Data System' intentions, beliefs or current expectations concerning, among other things, market prospects, growth, strategies, and the industry in which Samsung operates. By their nature, forward-looking statements involve risks and uncertainties, because they relate to events and depend on circumstances that may or may not occur in the future. Samsung cautions you that forward looking statements are not guarantees of future performance and that the actual developments of Samsung, the market, or industry in which Samsung operates may differ materially from those made or suggested by the forward-looking statements contained in this presentation or in the accompanying oral statements. In addition, even if the information contained herein or the oral statements are shown to be accurate, those developments may not be indicative developments in future periods.Logos remain the property of their respective owners. So there!

4

ClassicITusuallystartshere…

Copyright © 2015 Samsung SDS Co., Ltd. All rights reserved 5

ContainersFuelingPaceofReinven4on“TheNewIT”

Containersallowdevstobuildcomplex,distributedappsandservicesontheirnotebook,whichcanthenrun,unmodified,onpublicorprivateinfrastructure

•  Theyprovideisolation•  Theyarehighlyefficient(lowmemoryfootprint,lowstartuptime,rapidscaleup/down–often10Xoverallimprovement)

Anentireecosystemisrapidlybuilding…• Operationsautomation•  Lifecyclemanagement• Homogenous,super-efficientdatacenters

DisrupCveEffectsonClassicIT

•  LowEntryBarrier•  ServerlessStartups•  ShadowIT

•  Deprecatesexis4ngtechnologies•  Sharedblockstorage(SANs)•  Sharedfilesystems(NAS)•  Legacynetworking,e.g.h/wfirewalls

•  LessLabor(TinierEmpire!!!)•  Automateddeployment,scaling,repair,teardown

•  Cheaperserversandnetworkh/w•  Strictlyregimented,homogenous•  CaQle,notpets.

•  Keytechnologiesopensourced


ContainerUseMovingToEnterprises…fast!

• Usagemovingfromthebigserviceproviders(Google,Twitter,etc.)toenterprises.

• Evenverytraditionalenterprisesareadoptingthenewtechnology-CiscoITusingRedHatOpenshift,KubernetesandContainersforproductionworkloads.

DANGER!

ThesecompaniesareNOTjustmovingtoDocker!Theyarereworkingsystemsarchitectures,enablingsmallproducCveteams,automaCngoperaCons,sharingclusters,andautomaCngtesCng,i.e.:Movingtomicroservicesanddevopsapproaches.DOCKER(andcontainerorchestra4on)arekeyenablers,nottheendgame.

TheWideningGap

DedicatedPhysicalServers

FullVirtualizaCon

ContainerizaCon

Docker

OrchestraCon

AdHoc

Waterfall

Agile,Monolithic

Agile,Microservices

AdHoc

Epochal

ConCnuousIntegraCon

ConCnuousDeployment

ConCnuousQAChaosMonkey

50Servers/Op

40,000Servers/Op

DedicatedServers

DedicatedClusters

SharedClusters

“OldSchool”BigIT

Google,AWS,FB

1.6PUEDCs

1.1PUEDCsCloudInfra

(orjustusepublic!)

EnergyRecovery

Licensed&In-HouseSW

Lurk&ForkOpenSource

OpenSource

Contribute&LeadOpenSource

Copyright©2016SamsungSDSAllrightsreserved|ConfidenCal 12 15

Source:AdrianCockroghQp://www.slideshare.net/adriancockcrog/dockercon-state-of-the-art-in-microservices/3

MonolithicComplexDependencies

GoldPlated,ReliableHardwareExpensiveProprietaryDesigntoPreventFailure

Large,CentrallyControlledTeamsMonthstoProducConWaterfallProcessPonderous,ManualQAQualityIssuesCostly

MicroservicesLotsofSimple,IndependentParts

Small,IndependentTeamsConCnuousIntegraConConCnuousDeploymentChangesMadeinMinutesHighQualityUbiquitousAutomaConEfficient

CommodityHardwareSogwareReliabilityOpenSourceDesignedtoFailChaosMonkey

ClassicIT TheNewBar

What’saMicroservice???

•  Not:10-100linesofcode.AckPhQQ!•  Looselycoupledarchitecture•  Piecesindependently…

– Upgradeable– Scalable– Developed

•  Ifyouhavetoupgradeeverythingyouaredoingitwrong.

•  Mostly,singleteam(oreven,dev)


OrgImpacts

•  Microservices=Reorg•  Devops=Reorg•  Highpremiumonproductmanagement•  Highpremiumondevopsengineering,pipeline,andinternalplanorm


ThoughtProcess…MicroservicesapproachiscriCcalContainersareakeyenablerDevopsapproachiscriCcalContainersareakeyenablerMicroservicesandDevopsrequireadifferentorgWeshouldbetourfutureonthefutureWe’dliketohelplegacyservicesmoveforward

StrategicOutline:ProvideconsulCnghelpfororgissuesDesignanewcontainercentricdatacenterAddsupportforlegacy-styleapplicaCons

Automated,Container-CentricDatacenter

TheService

Containers

“Model”(Howtheapplicationshouldbedeployed,connected,protected,scaled,repairedandtorndown)

Servers

Network&StorageServices

Here’safuturethatsucks…•  DeploysomeOpenstack•  DeploysomeVMware•  DeploysomeMesosforHadoopandSpark•  DeploysomeCloudFoundryonOpenstack•  DeploysomeCloudFoundryonmetal•  UseAWSbutgetlockedintotheirveryniceservices•  UsesomeGCE•  DeploydedicatedUnix/Windows/Linux•  Throwswarm,kubernetes,GKE,andECSintothemixforgoodmeasure.RunsomeonMesos,someonOpenstack.

•  RunsomeMesosonOpenstackforgoodmeasure.


Moresuck…•  BuildexperCseonallofthosesystems…$$$,ifyoucanevendoit.

•  Paylotsforsupportsinceyouhavetoomanydifferentdistributedsystems

•  Trytofigureouthowtolog,alarm,debugallofthat.

•  Trytofigureouthowsecureallofthat•  Trytomanagecomplexitybyassigningeachsystemit’sowndedicatedcluster

•  Enjoy<10%systemuClizaCon•  HaveunsaCsfiedcustomersduetoreliabilityandperformanceissues.


•  70%DatacenteruClizaCon

•  MulCplereleasesaday/CD

•  10,000-40,000serversperadmin

•  DestrucCvetesCnginproducCon

•  No-ops(Selfservicedeployment)

•  Goveryfastandbreaknothing

•  Commodityhardware

•  Extremelylowlicensingspend(opensource)

•  ComprehensiveAutomaCon

•  ExtensiveAnalyCcs

HighBar-IndustrialGradeCompute

#GIFEE

Yeah,wewantthat!


ControlandRiskRiskreducConisaboutcontrolVendorlock-inislossofcontrolMorecontrolleadstolowercost

OpenSource+SamsungEngineers=Control

Goal…

23

Standardize,Containerize,Deploy

…toSamsungDataCenters.

…todevelopersystemsforagilityandproducCvity.

…topublicvirtualmachineclouds.…tonewandevenmoreefficientpubliccontainerclouds.

OpenSourceContainerizaCon•  BreakingvendorlocktoexpensiveproprietarysoluCons•  MoreefficiencyindatacenteruClizaCon

•  Easierpathtocommoditycompute/whiteboxinfrastructure

•  Greaterportabilitytoandfrompubliccloudsystems

•  Moreagilityindevelopmentanddeployment

•  Rapiddevelopmentanddeployment

•  Deploymentfreedom/portability

•  Scaleup/Scaledown•  Lowestpossiblecostdeployments(capexandopex)

StrategicElements•  Thededicatedclusterisdead.•  Collorary:Weneedlargesharedclusters.•  OnlyonekindofturtleattheboQom

– AWSTurtlesorGoogleTurtles?–  StandardizaCon–  CommodiCzaCon– AutomaCon– Doonethingwell,keepitassimpleaspossible

•  NeedtosupportlegacyapplicaConporCngandcloudnaCve

•  Control:Opensource,nolock-in


ExecuCon


Target Data Center Stack

BareMetal

Container-opCmizedhost

ContainerMgr

ContainerOpCmizedClusteringandResourceAllocaCon

PaaSappsJava/Scala

RubyWebDev

Non-PaaSAppsBackendprocessing

LegacyApps3rdParty

PublicCloud BareMetal


ContainerMgr

PublicCloud BareMetal


ContainerMgr

PublicCloud

PaaSCluster

PaaSCluster

RelaConalDB

BigDataDB

AnalyCcsCluster

BigDataDB

AnalyCcsCluster

Apps

Apps

DatabasesandAnaly4csHadoop

Non-SQLDB(Cassandra,Mongo,etc)SQLDB(MySQL,Postgress,etc)

ContainerOpCmizedNetwork

ContainerOpCmized•  ContainerOpCmizedHosts

–  Verysmallfootprint–onlywhatisneededtoruncontainers–  VeryfastbooCng–  Veryfastupdate–securityrequiresup-to-dateinfrastructure–  Examples:–  CoreOS–  Atomic(Redhat)–  Snappy(Ubuntu)–  RancherOS

•  ContainerOpCmizedNetwork–  ExplosionofIPaddresses:IPAddresspercontainer–  IPV6

•  ContainerOpCmizedOrchestraCon–  Highrateofcontainerstart/stop–  Higherdensityperserver

“Opinionated”CatsAhead…


WrongTurtlesattheBoQom


Openstack

Kubernetes

UpsideDown


Kubernetes

Swarm

MuchMoreSensible


Swarm

Kubernetes

TurtlesStackedTooDeep


Openstack

Kubernetes

Mesos

KubernetesisnotaPaaS.

TypeIKubernetesUser:WantstomanageasingleappTypeIIKubernetesUser:Wantstomanagedatacenters

BeQer


Openstack

Kubernetes Mesos

BeQersCll…


Mesos

Openstack Kubernetes

Whereweareheaded…


TheVMs,too.Incontainers.

MesosOpenstackSwarm

Kubernetes

PaaS

StrategicConclusion


ThereCanBeOnlyOne*

*Foreachenterprise.IFyoureallywantGIFEE.

OpCon:Openstack•  MulCplesupportopCons–nosupportlockin.•  Wrongtypeofturtle•  BurdenedbyfeaturiCs

–  Collorary:Notkeepingitsimple

•  Lackofstrongtechnicalleadership•  Knownissuesatscale•  Toomuchreligiousfervor

– NobodyexpectsthePythonistas!•  Haveusedthisinproduc6on,opinionisqualified.


OpCon:Mesos•  Adecentchoice,proventrackrecordatscale•  Singlevendorcontributorecosystem

–  Feelslikelock-intoMesosphereifyouwantcommercialsupport

–  ConfusingposiConfromMesospherevsDCOS•  ComplexandneedingsignificantexperCsetoconfigure

operate–  …butbeQerthanopenstack!

•  Goodecosystemaroundbigdata(Spark,Hadoop)•  RootsinscienCficcompuCng,notmodernservice

development….–  Containernetworking–  Servicediscovery

•  Havenotusedthisinproduc6on.


OpCon:Swarm•  Notreallyinthesamecategoryastheseothersystems,butDockerseemsheadedthatway

•  Reallyadeploymenttool,notaclustermanagementsystem

•  Notsuitableforlargescale•  Singlevendordominated•  SupportforAPCC,RKT,etcseemschallenged.•  Verysmallcontributorbase•  Wouldn’tconsiderusingthisinproduc6onatscale,soopinionisunlikelytochangeforawhile.


Choice:Kubernetes•  Gooddesignbyavibrant,healthycommunity

–  MulC-contributor•  Builtfromscratchformodernservicedevelop/deploy

–  PodsaretherightabstracCon•  Rapidpaceofimprovement

–  BehindMesosatpointinCme,buthighervelocity–  Our“BestFuture”bet

•  MulC-vendorsupportopCons(nosupportlock-in)–  SCllnascent.

•  Greatcontributorswiththerightexperience•  BesthighscalepubliccloudcontaineropCon

–  GoogleContainerEngine•  SupportsmulCplecontainerspecs:DockerandAPCC

42

KubernetesConcerns•  CNCFistherightdirecCon,butslowge|ngofftheground.•  Featuri6sriskiseverpresent.•  Weareconcernedandwillcon6nuetopushaspriori6es:

–  HA–  Scaling–  Reliability–  OperaConstooling

•  AccessibilityfornewusersiscriCcaltoecosystemhealth:–  Quickstart–  DocumentaCon–  UIs

•  Lookingforwardtomorecompaniesinthedirectsupportecosystem.

43

AFewWordsAboutPaaS

•  WeneedtosupportmulCplePaaSframeworks.

•  WedonotwanttodedicateclusterstoeachPaaSsystem.– CloudFoundry/Diego/BOSH

•  PaaSworkloadneedstosharedwithotherworkloads

•  APaaSshouldnotbeinthecontainerschedulingbusiness!


AFewMoreWordsAboutPaaS

•  CloudFoundryneedstorebasetoKubernetes(andMesos)toliveinthisworld.– DitchBOSH– DitchDiego– Focusonvalueupthestackratherthandown.

•  KudostoDeis,ApacheStratos,andOpenShig


SDSARefArchitectureEvoluCon

CoreOS

Docker

Kubernetes1.1+

AWS

Flannel

Apps

Terraform

Ansible

1000NodeStackBaseline

OpCmizedOS

Container

OrchestraCon

Compute

Networking

Provisioning

SamsungKubernetesEngagement

•  Wearenotforking•  Wearepushingbackourworkeitherintokubernetesorintoourgithubrepo–Samsung-AG.

•  2016–moreworkonnetworkingandstorage,with1000nodeclustersasthebaseline


WhyareweinvolvedinK8Scale?

•  WewantverylargeclusterswithcrossapplicaConresourcesharing– AlsolookingforwardtoUbernetes

•  WebelievewecanmakeaposiCvecontribuContomakethishappenfasterandbeQer.

•  Webelieveweneeddeeptechnicalinvolvementtobuild/deploy/operateatscale.

•  Welearnbydoing.


[email protected]://github.com/Samsung-AGSeaQle,WA

SDSRA…

WorldClassExperCseinModernBestPracCces:OrgDesign,AutomaCon,Devops,Containers,DataCenters,OpenSource


data center design / microservices

Software