INSPIRIT- Riverbed- Data protection and Disaster Recovery

1

Disaster Recovery with Riverbed

2

© 2009 Riverbed Technology. Confidential. IMPORTANT NOTE: The roadmap is for information purposes only and is not a commitment, promise or legal obligation to deliver any new products, features or functionality. The development, release, and timing of any features or functionality described remains at Riverbed's sole discretion.

Too much data to protect… too important to neglect

DATA GROWTH: Total disk storage revenue (new capacity) grew 20.7% in Q2 1

74% manage more than 50 TBs of storage capacity 2

45-fold data growth by 2020 3

IMPORTANCE: 73% rank BC/DR as a top 2010 IT initiative priority 4

American Eagle Outfitters retail website outage for 8 days Virgin Blue has 21 hour reservation system outage, massive delays Facebook down for 2 ½ hours! OMG!

1 - IDC’s Worldwide Quarterly Disk Systems Tracker, Sept. 20102 - Gatepoint Research survey of 1,000 respondents, Sept. 20103 - IDC’s “The Digital Universe Decade – Are You Ready”, May 20104 - Wedbush, Wave Q4, Dec. 2009

3


Storage Challenges

The performance problem

Limited protection windows

Increased recovery SLAs

The growth problem

Ever-increasing data

Limited bandwidth

The protection problem

Move data offsite

Secure data offsite

Easy recovery for offsite data

The integration problem

How do I leverage existing systems?

How do I minimize disruption?

4


Tape versus network backup

Local tape backups• Requires expensive hardware

in every office• Heavy administration• Low recovery rates• Increased security risk for lost

or stolen tapes

Eliminate local tape backups• Reduces costs in every office• Lights-out administration• Much more reliable• Reduced security risk

… = $ ? X !!

5


Tiering storage versus comprehensive coverage

Protect only “mission critical” data• Hard to identify• Changes frequently• Unclear inter-dependencies• Limited coverage of enterprise data

Protect all enterprise data• Everything is ready to recover• Go live instantly when you need it

6


Adding bandwidth versus optimizing connections

= 2 x $$

Get more from existing WAN links to every site

• Control expenses• No project delays• Maximize current infrastructure

Purchase dedicated, high bandwidth WAN links to every site

• High ongoing expense• Delays DR project implementation• Inefficient utilization• More links to manage and monitor

7


WAN Optimization entering mainstream maturity for BC/DR

WAN optimization complements your existing backup and replication approaches, non-disruptive.

8

Steelhead WAN Optimization for DR

9


Comprehensive Steelhead Product Deployment9

WAN

PRIMARY DATA CENTER

(Replicate to Secondary Data Center)

BRANCH OFFICES

(Backup / Replicate to Primary Data Center)

MOBILE WORKERS

(Backup Laptops)

Steelhead Appliances(Clustered)

Steelhead Mobile

Steelhead Appliance

CascadeCentral Management ConsoleSteelhead Mobile Controller

Interceptor

Virtual Steelhead

Steelhead Appliances(Clustered)

SECONDARY DATA CENTER

(Recovery Location)

10


Unoptimized Limited bandwidth

• Due to prohibitive costs or capacity availability

• Constrains maximum throughput

• Limits rate of backup or replication

All the data, a fraction of the WAN traffic

Optimized Expanded WAN capacity

• Adaptive Data Streamlining delivers up to 50X more space

• Deduplicates redundant data from the WAN

• Compresses new data for transfer

11


Unoptimized High Latency

• Lowers throughput over long distances

• Sometimes congestion causes loss, which FEC won’t help

Bandwidth under-used

Filling the pipe , optimizing bandwidth

Optimized High throughput

• Transport Streamlining, MX-TCP, & HS-TCP

• Tune the network• Fill the whole pipe• Mitigate loss and retransmission

12


Unoptimized Traffic contention

• One network for both DR and normal business operations

• Congestion• Impacts both DR and user

applications

Balancing traffic between DR and business apps

DR trafficApp traffic

Optimized Prioritized sharing on common

network links• Quality of Service (QoS)• Guarantees performance• Gives you control

13


Riverbed works with ALL leading storage vendors

XPEVA

MDS

7500

S3

Tivoli Storage Manager NetBackup

Backup ExecPureDisk

SnapMirrorSnapVault

http://images.google.com/imgres?imgurl=http://bt.baskent.edu.tr/logos/MicrosoftLogo.jpg&imgrefurl=http://www.gaufire.blogspot.com/2008_06_01_archive.html&h=686&w=2846&sz=80&hl=en&start=123&tbnid=qXPGv2e2QgNLeM:&tbnh=36&tbnw=150&prev=/images?q=microsoft+Server+2003+logo+large&start=108&gbv=2&ndsp=18&hl=en&safe=off&sa=N

14


EMC Partnership

E-Lab qualified:• SRDF, Recover Point, Celerra Replicator, Mirrorview, SAN Copy

Unique SRDF and FCIP protocol optimizations Auto-configure for SRDF

A few of our joint customers:

15


IBM

Partner with IBM Global Technology Services Qualified with XIV Remote Mirroring Also great results with Global Mirror & TSM

83% bandwidth reduction


16


NetApp Partnership

FAS and V-series platforms SnapMirror replication & SnapVault backup Performance example:


17


Hitachi Data Systems (HDS)

Certified Hitachi Universal Replicator (HUR), True Copy, Hitachi NAS

Also a customer:

18


Dell EqualLogic

EqualLogic PS Series iSCSI SAN arrays Optimize Auto-Replication Joint ROBO initiative Performance example:

• Replication – 48 GB Exchange Replication - 20 Mbps• 100ms Latency –Time to Complete (in minutes)


19


What about business continuity?

Faster restore across the WAN Or go live in DR site

What happens in the case of a site disaster?• Flood, hurricane, winter storm, or maybe just closed due to the flu• An office is closed for some time…• How do you continue to function as a business?

Work from anywhere with Steelhead Mobile!

20

Cascade for DR

21


Monitoring the enterprise

Monitor application performance Resolve IT issues faster Consolidate IT assets cheaper and faster Analyze and optimize WAN traffic Detect security threats Increase IT staff productivity and reduce downtime Differentiated analytical capabilities, network drill-down, discovery & dependency mapping

Cascade

Cascade is an enterprise-wide network and application visibility solution

22


Top 3 Visibility Requirements for DR

What systems needto be protected?

Is the backup/replicationprocess working?

Has a disasterhappened?

Enterprise-wide discovery & dependency mapping

Application performance management

Detect outage, troubleshoot cause, start restore or failover sooner

Cascade – monitor DR operations, faster problem detection & resolution

23

Whitewater for DR

24


Backup & Disaster Recovery Today

Data Center

DR site

Replication

Back Up Today

BackupMedia Server

• Expensive – requires 2nd site• Backup & archive data

continuously growing• Steelheads optimize bandwidth

& replication, not disk

• Slow• Off-siting is tedious• Difficult to restore quickly• Risk of loss, theft in transport• Media obsolescence

Disk

25


Implementing Cloud Storage

BenefitsROI is great - 3x less costly

OpEx vs CapExGrowth on demand

ChallengesNeed to speak “cloud”

Requires changes (tech & process)

Latency kills performanceSecurity concerns

26


Cloud Storage

Cloud Service ProviderData Center

BackupServer

Faster restore

Optimized backup to cloudCIFS/ NFS

Encryption Deduplication Acceleration

Local Storage

Cloud API (e.g, REST)

Backup Tomorrow – Riverbed Whitewater

• 30-50% cost reduction• Faster backup and restore• Flexible data off-siting • Secure & protected• No change to backup process

Single-ended solution – only one box in the DC

27


Value Prop for Whitewater

Simple Deployment & Integration

Works with existing backup software

Flexible – Hardware or Virtual

Accelerated Backup & Recovery

Shorter Backup Windows

Quicker Restore Times

Optimized over the WAN

Secure data in motion & at rest

Encrypted data onsite, inflight and at rest

256-bit AES , SSLv3

Easy key management

Cut Storage Costs

Deduplicate cloud storage

Use cloud for pay-as-you-consume storage

28


Integrate with Your Existing Backup Tools

Riverbed Whitewater Technology Differentiators

Accelerate Backup and Recovery

Recovery in minutes instead of hours

Deduplicate Data Secure Data in the Cloud

Sub-block data reduction provides 10-

30x dedupe on average

Provide AES-256 encryption with keys managed from your

data center

No rip-and-replace:Work with your existing backup

software; co-exist with existing backup targets (tape & VTL)

29


Cloud Storage


BackupServer

0101010111011010

0101010111011010

0101010111011010

0101010111011010

0101010111011010

0101010111011010

0101010111011010

Uses existing backup tools 10-30x average dedupe Local restore for recent

data

Accelerated cloudbackup and restore

Works with cloud storage protocols

Limitless storage Elastic & low cost Optimized dedupe

How it Works : Accelerated Backup

Faster restore

Optimized backup to cloudCIFS/ NFS

30


Cloud Storage


BackupServer

0101010111011010

0101010111011010

0101010111011010

0101010111011010

Recent backups are still on appliance disk

Disk retains up to a month of backups

Recent restores are on the LAN

Older data is restored on demand from Cloud

“local-like” restore over the WAN

How it works : Local-like Restore

Faster restore

0101010111011111101100101010111011010

0101010111011010

0101010111011010

0101010111011010

01111 0110001

0110

From the customer perspective – it is just a simple local-like restore

32


Real world results

Enterprise coverage – G&T Conveyor• “Our massive […] DR jobs that basically didn’t work, even over four days, can now be done

successfully in a day.” Limited bandwidth – Cubist

• 55x faster replication of VMware virtual machine snaps from Milan to Boston on T1 ROI from eliminating backup – IDC study of Riverbed customers 1

• “The average annual savings on tapes, tape drives, and vaulting is $84,639.” RPO – Liz Claiborne

• “Steelhead appliances cut WAN-based backup windows from 16 hours to just 6 hours. This enables us to do more frequent, more reliable backups.”

RTO – IDC study of Riverbed customers 1

• “Restore time has been improved by 83%, on average, since the deployment.” Shared bandwidth capacity – AAR

• 8.2x capacity increase on OC-3 for EMC SRDF between Sydney and Melbourne• Avoided need for a dedicated 100 Mbps circuit

1 - “Improving the Business Value of WAN Optimization”, IDC, Jan. 2010

33


So now what?

Identify the primary goal of the project• Prohibitive bandwidth requirements and costs • Meeting data protection windows

Engage all IT teams to understand environment and requirements• Storage, network, server, applications, desktops

Design, validate, & prove Build your business case (examples)

• Downsizing the WAN link• Deferral of WAN upgrades• Improved RPO/RTO for business continuity goals• Balancing DR versus other apps on the WAN

34

Thank you

35

Appendix

36


Highly available WAN optimization

Choose the right level of availability Plus:

• Fail to wire or Fail to block• RAID data stores• Redundant power supplies• Fault tolerant SSDs• WCCP & PBR compatible• Synchronized data stores

No single point of failure

Interceptors

Serial

Parallel

37


Scalable Range of Models

Steelhead appliances scale with upgradable families to fit any environment

• Branch

• Mid-sized

• Data center

{{{

Model Bandwidth Connections Data Store

Steelhead appliances (Mbps) (Optimized) (GB)250-L 1 30 40250-M 1 125 40250-H 2 200 40550-M 2 300 80550-H 4 600 801050-L 8 800 1001050-M 10 1300 1001050-H 20 2300 2002050-L 45 2500 4002050-M 45 4000 4002050-H 45 6000 4005050-L 90 7500 6005050-M 90 10000 6005050-H 155 18000 8006050- 310 50000 35007050-L 1000 75000 22007050-H 1000 100000 4400Interceptor 9350 12000 1000000

* Includes all SSD data store & fault tolerance for highest performance

**

38


Security features

Fits your requirements FIPS 140-2 CC EAL4+ JITC approved

•SSL (or IPSEC) between Steelheads•SSL & HTTPS acceleration•Encrypted data store

•Up to 256-bit AES encryption•Role-based administration•Improved SMB signing support•Encrypted MAPI & Notes support•Steelhead Mobile VPN interoperability•Unified Trust Model

Access to managementconsole

39


Whitewater scales to meet your requirements

Virtual WW WWA-1000 WWA-2000

Raw Local Disk Capacity

2 TB 8 TB 16 TB

Logical Capacity1 40TB 160 TB 320 TBIngest Rate 0.25 TB/hr 1 TB / hr 1 TB / hourAverage Data Reduction2 10-30x

1 Logical capacity refers to the amount of deduped data that is stored on disk. This can vary based on the type of data being backed up. For illustration purposes, a dedup factor of 20x has been considered

2 Reduction can be expected up to the numbers listed above. Data reduction varies based on many factors including the type of data and frequency of backup.

40


Compatibility Matrix for Whitewater v1.0

Backup Products

Cloud Storage

With more to come

NetbackupBackup Exec

Synaptic Storage as a Service

Atmos

41


Enterprises Testing Whitewater

Unified Communications

Manufacturing

Engineering Consulting Services

Natural Resources

42


Customer Results Customer:

• Cubist Pharmaceuticals• Publicly traded pharmaceutical company

Environment:• Milan, Italy connects to Boston, USA data center via a Cisco VPN client• Using VMware Consolidated Backup (VCB)

Problem:• How to protect a 90% virtualized Windows and Linux server environment

Solution:• Replicating tens of gigabytes of VMware snapshots to DR site daily• Up to 55x improvement in replication time for virtual server infrastructure

“The Steelhead products have given us unbelievable performance when replicating VMware. We now see upwards of 50 to 55x improvement in data transfer for a snap of a VMware machine.”

- Mike Geldart, Sr. Manager, Computer Operations

43



• Allens Arthur Robison• A law firm with 1500 staff in 14 cities across Australia and Asia

Environment:• EMC SRDF/A with GigE interfaces on Symmetrix DMX

Problem:• Analysis suggested an expensive and dedicated 100Mb/s

link between Sydney and Melbourne, • Had only a 20Mb/s link shared with their usual WAN traffic

(Email, file sharing, etc.) Solution:

• Reduced link consumption by 8x• “Deliver the impossible” – increase DR capability without

increasing bandwidth or impacting application performance

44


Customer Result Customer:

• Operates much of a very big state’s power grid Environment:

• Redundant OC-3 with 12 msec RTT• SRDF/A with other apps• Oracle database replication

Problem:• Backup link was also being used to meet existing data transfer

needs.• Needed to reduce the impact of replication traffic on WAN and other

critical applications Solution:

• Reduced link consumption by 6.5x• Enable true network redundancy enabling moving all traffic on the

primary link• Four 6020 Steelheads deployed

45



• A leading credit card processor for merchants Environment:

• Redundant OC3 with 45 msec RTT• Oracle database replication

Problem:• Needed to reduce the impact of replication traffic

on WAN and other critical applications Solution:

• Reduced link consumption by 6x• Avoided bandwidth upgrade• Four 6020 Steelhead appliances deployed

46


360 Architects

Background• Easily replicating over 200GB of data per day from Kansas City to Dallas.• Using native EMC Celerra V2 Replication, and EMC Centera replication.

Results• Seeing 85% to 98% optimization rates for the replication traffic. • Bandwidth capacity increase is currently over 7.07X for the 6Mb fiber MPLS connection.

Key Benefits from Deploying Riverbed• Meeting data protection goals• More comprehensive DR with current bandwidth

“Guys, this is ridiculously good performance and well beyond my expectations and planning numbers. I feel like the more replication traffic we throw at the Steelheads, the better the performance will be. “

Jeff Hammer,

INSPIRIT- Riverbed- Data protection and Disaster Recovery

Documents

offsite data

new products

important note

riverbed technology

legal obligation

information purposes

secure data

riverbeds sole discretion