ditas project presentation v1.0

28
DITAS Project Introduction

Upload: oliver-barreto-rodriguez

Post on 22-Jan-2018

41 views

Category:

Technology


0 download

TRANSCRIPT

Page 1: Ditas Project Presentation v1.0

DITAS Project

Introduction

Page 2: Ditas Project Presentation v1.0

Context

Page 3: Ditas Project Presentation v1.0

Data-Intensive Applications

Easy to implement

scalable solutions

Data management is

totally demanded to the

cloud providers

Robust solutions

Latency could be

significant

No control over data

management

(security/privacy

compliance checking)

Total control over the

data management

Data latency reduced

Higher investments

No scalability

on Cloud on the Edge

Page 4: Ditas Project Presentation v1.0

Mixing the two

Cloud On premise

Scalability

Reliability

Latency

Security

Compliance

Scalability

Reliability

Latency

Security

Compliance

DITAS

Page 5: Ditas Project Presentation v1.0

DITAS Objectives

1. To improve the productivity of developers

2. To enhance the data management in mixed cloud/fog

environment

3. To improve the execution of data-intensive applications

through data movement strategies

4. To provide an innovative execution environment for data-

intensive applications

5. To maximize the impact on the market of developers and

adopters of data-intensive applications

Page 6: Ditas Project Presentation v1.0

What is a DIA?

We call an application data-intensive if data is

its primary challenge:

the quantity of data,

the complexity of data,

or the speed at which it is changing [5]

Two perspectives:

Data production

Data consumption

Page 7: Ditas Project Presentation v1.0

Our position

Source [5]

Page 8: Ditas Project Presentation v1.0

Our case studies

Characteristic Industry 4.0 e-Health

Data size Depending on the sensors attached to

the machine, it generates 100MB –

600MB per day

GigaByte if characters, Terabyte if

images

Complexity of analysis TBD Taxonomy problems and source

hetherogeneity increase the analysis

complexity

Heterogeneity of data Low High

Number of data sources Between 1 and 3 Several , depending upon the

domain; on average 5 to 7 sources

Distribution of data Not distributed Depending upon the localization of

the data sources. For multicentric

clinical studies the sources may be

located in different , geographically

distinct institutions

Timeliness of processing Needs can vary from near real-time to

hours depending on the

service/application

No online query are required,

therefore the response time is not an

issue for complex analysis.

Page 9: Ditas Project Presentation v1.0

s

s

s

ss

s

Our model

Store

Acquire

Consume

Dismiss

help you on your daily tasks on Data

Management

Page 10: Ditas Project Presentation v1.0

Our model

Data sources

Data format

Transmission

protocols

s

s

s

ss

sStore

Acquire

Consume

Dismiss

help you on your daily tasks on Data

Management

Page 11: Ditas Project Presentation v1.0

Our model

Where to store

DFS with IoT

Security/privacy

s

s

s

ss

sStore

Acquire

Consume

Dismiss

help you on your daily tasks on Data

Management

Page 12: Ditas Project Presentation v1.0

Our model

Data analytics /

operational processes

Lambda/Kappa

architecture

s

s

s

ss

sStore

Acquire

Consume

Dismiss

help you on your daily tasks on Data

Management

Page 13: Ditas Project Presentation v1.0

Our model

Delete

Freezing

s

s

s

ss

sStore

Acquire

Consume

Dismiss

help you on your daily tasks on Data

Management

Page 14: Ditas Project Presentation v1.0

Dismiss

Our model

Data movement

is covered in all

the phasesStore

Acquire

Consume

Dismiss

help you on your daily tasks on Data

Management

Page 15: Ditas Project Presentation v1.0

About the results

Page 16: Ditas Project Presentation v1.0

Expected results

DITAS-SDK

To design and implement data-intensive applications

DITAS Execution Environment (DITAS-EE)

To run data-intensive applications where data movement is

working behind the stage

Data and computation movement techniques

To satisfy the developer requirements

Strategies for application improvement

To be sure that data movement strategies are properly

working

Page 17: Ditas Project Presentation v1.0

Solution at a glance

DITAS

SDK

Device and cloud manager

Identity manager

Application deployment

manager

DITAS Execution

EnvironmentDITAS Execution

Environment

DITAS Execution

Environment

DITAS Execution

Environment

Page 18: Ditas Project Presentation v1.0

DITAS as a service…

DITAS developer

PaaS

(design) (execution)

IaaS• Hide the complexity of the infrastructure

• Where the application is deployed

• What the developer uses

• Supports the design of DIA

• Enable the data movement

Page 19: Ditas Project Presentation v1.0

IaaS in DITAS

IaaS

Common Accessible Framework

Virtual access to

data in the

federated storedData in the cloud store

Page 20: Ditas Project Presentation v1.0

PaaS

PaaS in DITAS (Design) -

SDK

Data Utility definitionData Security and

Privacy

Virtual Data Container Definition

Application Profiling and Deployment strategies)

Page 21: Ditas Project Presentation v1.0

PaaS

PaaS in DITAS (Execution

Environemnt – DitasEE)

Data movement

strategies

Data analytics

Execution Engine

Data movement enactor

Data monitoring

Page 22: Ditas Project Presentation v1.0

About the work plan

Page 23: Ditas Project Presentation v1.0

Milestones

General architecture

design and case studies

refinement

Detailed component architecture

First releaseFinal release and mature

market analysis

Final validation, joint

sustainability, and established

sustainability body

M6MS1 M12MS2 M18MS3 M30MS4 M36MS5

WP1 WP2

WP3

WP4

WP1

WP2

WP3

WP4

WP5

WP2

WP3

WP4

WP5

WP5

Iteration 1 Iteration 2

Page 24: Ditas Project Presentation v1.0

How DITAS can help your

workflow

Page 25: Ditas Project Presentation v1.0

Phase 1 – system knowledge

Device profiling

How to access

What is able to do

Data profiling

Which are the data

consumed/produced

Characteristics of the

data IaaS

Page 26: Ditas Project Presentation v1.0

Phase 2 - development

Developer defines the data-processing pipeline (process) Data are seen as VDC

Agnostic from the platform (unless data does not exist)

Developer defines the QoS At which granularity

(process/function)

DITAS

SDK

IaaS

PaaS

Page 27: Ditas Project Presentation v1.0

Phase 3 – deployment

Based on the system system knowledge functions are properly deployed

Balance between cost/benefits deployment

Questions: Do all the nodes know

the complete process?

Do we need to have a central node?

Application deployment

manager

DITAS Execution

EnvironmentDITAS Execution

Environment

Page 28: Ditas Project Presentation v1.0

Phase 3 – execution

(adaptive)

When data processing is in place data and computation can be moved Move data among

devices

Move computation among devices

Monitoring Centralized/distributed

Application deployment

manager

DITAS Execution

EnvironmentDITAS Execution

Environment