EMC Storage Strategy · HDFS / Object Storage. Eventual Consistency. ... Architecture In-Depth. Isilon. vSAN. ... HUS VM. NetApp. FAS . EMC. ScaleIO. VNX . VBLOCK.

1© Copyright 2014 EMC Corporation. All rights reserved.© Copyright 2014 EMC Corporation. All rights reserved.

EMC Storage StrategyMuhammad Salama

Advisory Systems Engineer – EMC Gulf


Enterprises Are Being Redefined By Software


BUSINESS


INFRASTRUCTURE

I think the pic is maybe more of a hybrid cloud thing

(maybe label it as well)

HYBRID CLOUD


Workload Diversification Drives Infrastructure Innovation


It’s a Natural Principle!


Mainframe, Mini ComputerTerminals

LAN/Internet Client/ServerPC

Mobile Cloud Big Data SocialMobile Devices

MILLIONS OF USERS

THOUSANDSOF APPS

HUNDREDS OF MILLIONS OF USERS

TENS OF THOUSANDSOF APPS

BILLIONSOF USERS

MILLIONSOF APPS

Source: IDC, 2012


Platform Application Characteristics

HDFS / Object Storage

Eventual Consistency

Tolerance For Data Loss

Application-Based Resiliency

Block / File Storage

Transactional Consistency

Little Tolerance For Data Loss

Infrastructure-Based Resiliency

LAN/Internet Client/ServerPC


700%By 2016

70%By 2016

Source: IDC, 2012


Platform Application Characteristics

HDFS / Object Storage

Eventual Consistency

Tolerance For Data Loss

Application-Based Resiliency

Block / File Storage

Transactional Consistency

Little Tolerance For Data Loss

Infrastructure-Based ResiliencyLAN/Internet Client/Server


700%By 2016

70%By 2016

Source: IDC, 2012


How has this been driving ourPortfolio Strategy


ADVANCEDDATA SERVICES

SIMPLEDATA SERVICES

Workloads Have Different Requirements

TransactionProcessing

OnlineArchive

VM Infrastructure

DecisionSupportFull Motion

Video

HighPerformanceComputing

WebApps

ContentManagement

ITApplications

StorageAs-A-Service

HomeDirectories

Email

ApplicationDevelopment

TechApps


This Requires a Diverse Storage Portfolio


SIMPLEDATA SERVICES

VMAX

VNX

Isilon

XtremIO

AtmosData Domain

Avamar & NetWorker

XtremSF



The Four Storage ArchitecturesTYPE

1ClusteredScale Up & Down

• Dual Controller• Shared Meta-Data• Transactional Commits• General Purpose Storage

TYPE

3Loosely CoupledScale Out

• Independent Controllers• Distributed Meta-Data• Transactional Commits• Large Content Repositories

TYPE

2Tightly CoupledScale Out

• Multi-Controller Grid• Shared Meta-Data• Transactional Commits• Mission Critical Workloads

TYPE

4DistributedShare Nothing

• Independent Controllers• Distributed Meta-Data• Non-Transactional Commits• 3rd Platform Applications


Architecture In-Depth

0

1

2

3

4

5$/IOps

$/GB

Latency

AggregatePerformance

QoSRAS

Scales Down

Manageabilityand Scale Out

SolutionIntegration

Available (five 9’s)

Cost Efficient

Broad Integration

Scale Out

Aggregate Perf

VNX

TYPE


• Dual Controller• Shared Meta-Data• Transactional Commits• General Purpose Storage


Architecture In-DepthTYPE


• Multi-Controller Grid• Shared Meta-Data• Transactional Commits• Mission Critical Workloads

0

1

2

3

4

5$/IOps

$/GB

Latency


QoSRAS

Scales Down


SolutionIntegration

IsilonVMAX

XtremIO

Dozens of Nodes

Predictable Latency

Very High Reliability

Cost of Entry

$/IO & $/GB


TYPE


• Independent Controllers• Distributed Meta-Data• Transactional Commits• Large Content Repositories

0

1

2

3

4

5$/IOps

$/GB

Latency


QoSRAS

Scales Down


SolutionIntegration

Architecture In-Depth

Isilon

v S A N

Aggregate Perf

Hundreds of Nodes

Low $/GB

Latency

Entry Point Price


Architecture In-DepthTYPE



0

1

2

3

4

5$/IOps

$/GB

Latency


QoSRAS

Scales Down


SolutionIntegration

Incremental Scale

Simple/Invisible

Low Cost of Entry

Data Services

Ent Availability


The Four Storage ArchitecturesTYPE


• Dual Controller• Sharedx Meta-Data• Transactional Commits• General Purpose Storage• Transactional Commits

TYPE



TYPE


• Multi-Controller Grid• Shared Meta-Data• Transactional Commits• Mission Critical Workloads• Transactional Commits

TYPE


• Independent Controllers• Distributed Meta-Data• Transactional Commits• Large Content Repositories• Transactional Commits

Federation Layer


How do we apply this knowledge toApplication Workloads



SIMPLEDATA SERVICES

Workload Drive Architectures Not Vice Versa

DualController

Scale Up/Down

TightlyCoupled

Scale-Out

TightlyCoupled

Scale-Out

LooselyCoupled

Scale-Out LooselyCoupled

Scale-OutSharedNothing

SharedNothing


How Do We Simplify The Management And Operations

Of These Diverse Systems


Expanded Support : Native & OpenStack

ORACLEZFS

NetAppE-Series Solidfire

IBMXIV

IBMSVC

IBMDS8000

NATIVE INTEGRATION

Isilon

VMAX

HitachiVSP

(Block)

HitachiHUS VM

NetAppFAS

EMCScaleIO

VNX VBLOCK HP 3PAR HPLeft-Hand

DellEqual-Logic

XtremIO

VIPRController


Integrates With OpenStack & Microsoft Too

ViPR Data Services

VIPRController

Commodity Hardware Storage Arrays

VNX Isilon XIO3rd Party

VMAX


Where can you find out more?

www.emc.com/ppg

EMC Product Positioning Guide

http://www.emc.com/ppg

EMC Storage Strategy · HDFS / Object Storage. Eventual Consistency. ... Architecture In-Depth. Isilon. vSAN. ... HUS VM. NetApp. FAS . EMC. ScaleIO. VNX . VBLOCK.

Documents