A8: Improving Recovery Objectives though Integrated Remote Replication and Continuous Data Protection

1
A8 Improving Recovery Objectives though
Integrated Remote Replication and Continuous Data
Protection

• Gary Archer
• Senior Manager, Cross Platform Software Marketing
• EMC Corporation

2
Abstract

• Improving Recovery Objectives though Integrated
  Remote Replication and Continuous Data Protection
  
• Protecting data for mission critical applications
  has become a top priority for corporations
  seeking cost effective methods to recover from a
  number of different disaster scenarios. 
• In next generation data centers, whether a
  disaster is natural or man-made, loss of critical
  data is never acceptable, so CIOs are challenged
  to address these key business issues while
  maintaining fiscal responsibility. 
• At the same time, the pressure to minimize data
  loss and protect a broad range of applications
  becomes even more difficult when managing
  heterogeneous storage and server architectures. 
• This presentation will demonstrate how to use
  integrated Remote Replication and Continuous Data
  Protection (CDP) solutions to meet strict
  recovery objectives using fabric-based data
  protection solutions across a variety of
  applications and storage infrastructure. 

3
Outline

• Current Challenges/Requirements
• Introduction to Remote Replication and CDP
• Evolving data protection requirements
• Merging operational and disaster recoveries
• Fabric-based data protection
• Federated application protection
• Additional Resources

4
The CIOs Information-Storage and -Management
Requirements
The right local- or remote-replication solution
will help you overcome these challenges
5
Replication pain-points
Application response time
Local site
Remote site
Application-consistent recovery
Corruption protection
SAN
SAN
SAN
Disaster-recovery testing
Communicationscost
Existinginfrastructure
6
Local and Remote Replication Benefits

• Protect against local and regional site
  disruptions
• Continuous data availability
• Remote-recovery sites
• Meet regulatory requirements
• Support multiple service levels with tiered
  storage
• Provide near-instant recovery
• Migrate, consolidate, or distribute data across
  heterogeneous storage platforms
• Data center consolidations
• Technology refreshes
• Enable non-stop operations
• Application restart across volumes or arrays

7
Protection Based on Recovery Objectives (SLAs)
Secs
Days
Mins
Secs
Mins
Hrs
Hrs
Days
Years
Recovery Point
Recovery Time
Protection Methods
Recovery Methods

Synthetic Backup Real Time Replication
Capture on Write
Tape Backups
Tape Restores
Instant Recovery
Disk Restores
Disk Backups
Vaults
Point-in-Time
Archival
Snapshots
Roll Back
Surgical Search Retrieve
Enabling Technologies
Tape Automation
Continuous Data Protection
De-duplication
Remote Replication
Content Indexed Archival
Source SNIA Trends in Data Protection CDP and
VTL
8
Introduction to CDP

• What
• Capture every change as it occurs
• Protected copy in a secondary location
• Recover to any point in time
• How
• Block-based
• File-based
• Application-based
• Why
• Implementations of CDP can deliver near zero data
  loss with no impact to production and application
  consistent recovery. CDP customers can protect
  all data at all times and recover directly to any
  point in time.

Source SNIA Trends in Data Protection CDP and
VTL
9
Protection Technology Comparisons
CDP (RPOevery write)
High Frequency Snapshots (RPO15 mins)
Snapshots (RPO8 hrs)
Tape Backup (RPO24 hrs)
Time
10
Recovery Technology Comparisons
Applications Restarted
Last good image
Restart
Analyze
Restore
Traditional
Data LossltRPO
Application Downtime
Time
Restore time varies depending on technology
10TB from tape can take 12 hours
11
Local and Remote Replication with CDP
Continuous Data Protection
Application / Database Servers
Engine
OR Repository
Email / File Servers

• Local Operational Recovery
• Recovery is by application local
• Minimizing RPORTO is critical
• Retain shorter history (days)
• Individual application recovery
• Multiple applications / multiple CGs

12
Best Practices for CDP and Replication

• Classify an application data and capture SLAs
• Content grouping, retention policies
• Understand consistency requirements
• System-level, transaction, crash
• Capture storage requirements
• volume of data, change rates, history retention
• Assess communication requirements
• Bandwidth, costs,
• Be selective in your implementation
• Not all application data needs CDP technologies
• Use as primary technology to reduce backup window
• Not a replacement for traditional B2D or B2T
• Still take advantage of tape as backup target
  where appropriate
• Build out to enable addition quick restores
• Leverage combined technologies as required
• Single message, mailbox, table, file, folder,
  and/or volume
• Avoid lock-in and fork-lift upgrades
• Look to heterogeneous solutions
• Understand application, server, network
  transparency

13
Understanding Data Consistency

• Applications and data are interrelated
  (federated)
• All data movement must be stopped/started at the
  same point in time
• To restart applications you must have all the
  datanot parts of it
• Recovery requires dependent-write consistency
  across all volumes and systems

CRM
Order Entry
DB
DB
DB
SCM
Systems share informationhow do you get a
consistent view?
14
Multiple Consistency Groups

• Allows application recovery to be tiered by
  service level
• Multiple volumes per group
• Mixed RPOs within same infrastructure
• Provides independent replication controls
• Recover by group
• Start/Stop by group
• Optimizing by group
• Priority
• Compression
• Resource usage

OE
Target
Group 1
CRM
Target
Group 2
SCM
Target
E-mail
Target
Group 3
15
Consistency for multi-host applications
1 Linux (Web OE)

• Each tier has different SLAs
• Consistency groups per tier
• Operational Recovery of tier
• Parallel consistency across tiers
• Enable recovery to known point
• Disaster Recovery for tier
• Spans OS, applications, storage, servers
• Enables advanced functions
• Full environment cloning
• Application upgrade testing
• Data mining

Consistency Group
2 Windows (CRM)
CG
CG
3 UNIX (SCM, Financials, )
16
Using CDP for consistency for multi-host
applications
Customer X added
Customer Y added
Receivable X1 added
1
3
5
4
6
Order X1 received
Order Y1 received
Receivable Y1 received
2
Biz App
Order entry Web OE
Customer Records SQL
Supply Chain Oracle
CDP Repository
1
2
3
1
3
5
6
4
5
6
4
2
Mid-tier Storage
High-end Storage
17
Leveraging Intelligent Switches
Inside the Intelligent Switch
I/O Targets
Engine

Input I/O stream
Storage Target
Heterogeneous Hosts
Intelligent switch becomes storage target

• Intelligent Switches
• Fibre Channel switches with custom hardware for
  enhanced processing
• Capable of performing operations on data streams
  at ASIC speeds
• Controlled by instructions from external
  management software over FC (via APIs)
• The switch becomes the storage target for the
  hosts

18
Local Replication Process
1. Data is split original to target, copy to
replication engine
2a. Host splitter
3. Writes are acknowledged back from the
replication engine
2b. Intelligent fabric splitter
4. The appliance writes data to the repository,
along with timestamp and application-specific
bookmarks
Production volumes
Replica volumes
Repository
5. Write-order-consistent data is distributed to
the replica volumes
19
Remote Replication Process
1. Data is split original to target, copy to
replication engine
2a. Host splitter
7. Data is written to the repository
3. Writes are acknowledged back from the
replication engine
6. Data is received, uncompressed, sequenced, and
checksummed
2b. Fabric splitter
5. Data is sequenced, checksummed, compressed,
and replicated to the remote site
4. Engine functions FC-IP conversion
Data reduction and compression
Replication Monitoring and management
/ A
/ C
/ B
Local site
Remote site
Repository
8. Consistent data is distributed to the remote
volumes
20
Use Cases

• Exchange Remote Replication
• Database Remote Replication
• Local Application Upgrades using CDP

21
Use Case Exchange Remote ReplicationCustomer
Large global legal firm, Asynchronous Replication

• Remote Exchange Replication
• Overview
• 6 TB of production capacity (CX700)
• Replicating across 500-mile WAN link to DR
  location (CX500)
• Pain Points
• 15-minute RPO, quarterly DR tests, cross-host
  consistency
• Previously used host-based replication,
  performance issues
• Implementation
• Continuous replication, 1-minute lag, hourly
  application consistency
• Storage Groups Logs into Consistency Group
  (EX1)
• Leveraged Target Side Processing to test crash
  consistent images

22
Exchange Disaster Recovery
Capture Application Bookmarks (2) Periodically
put Exchange into backup mode - Close Image,
tag as BOOKMARK 1,2 - When tagging
complete, resume operation
Exchange 2003 Recovery Servers
Exchange 2003 Cluster
Test and Recover (3) Pause Image, mount
replicated volume to recovery server in TSP Mode
Initialization/Transfer (1) Consistency Group
Storage Groups Logs - Configure CG Journal -
Lag Policy 60 Seconds - Synchronize remote site -
Begin image capture
Engine
Engine
Deltas transferred to remote site
Journal Entries
IMAGE 121 BACKUP BOOKMARK 2 IMAGE 120
IMAGES 61 BACKUP BOOKMARK 1 IMAGE 60
IMAGE 1
Production
Disaster Recovery Site
23
Example Use CaseCustomer Mid-size Bank,
Asynchronous Replication

• Remote Site Database Recovery
• Overview
• Progress and SQL production databases (CX500)
• Replicating across 3 Mb/s link to remote DR
  location (CX500)
• Pain Points
• 60-minute RPO, quarterly DR tests, narrow WAN
  link (2 DS1s)
• Federate across multiple database systems
• Other solutions wouldnt work on existing WAN
• Implementation
• Continuous replication, 5-minute lag policy, WAN
  compression
• Federated database tagging, cross-group
  consistency
• Leveraged Target Side Processing for DR tests

24
Remote Site Database Recovery
Data Center
Remote DR Site
Database Instances
Database Instances
Compression used to reduce BW
Engine
Engine
Independently recover and test data from remote
site
Group Sets Cross-tagging for Progress and SQL
databases
SQL Repository
Next Slide
25
Cross-Tagging Federating Independent Databases

• SQL Remote Journal
• SQL VDI Checkpoints
• Federated Bookmarks Group Sets provide
  consistency across SQL and Progress DBs

26
Example Use CaseCustomer Leading Materials
Manufacturer, Local ReplicationCDP

• Customer Application (SAP) Upgrade
• Overview
• Major upgrade of SAP ERP and CRM systems
• Requires fallback strategy for upgrade on
  production
• Pain Points
• Test production image in test/dev lab
• DMX arrays in production, CX in test/dev
• Requires iterative test upgrades to streamline
  process
• Implementation
• Local replication with continuous data protection
• Replicate full copy of SAP systems to test/dev
• Leverage Target Side Processing to refresh
  test/dev image

27
Use Case SAP Upgrade with CDP
SAN Cloning of CDP images updates array LUNs to
track changes
Production
Test/Development
ERP
CRM
ERP
CRM
Engine
CDP image replicated to test SAN
ERP Test
ERP Production
CRM Test
CRM Production
28
Summary

• One protection technology can be used for local
  operational recovery and remote disaster
  recovery.
• To provide protection against both loss and
  corruption, replication must integrate CDP
  technologies.
• Intelligent fabric-based data protection is
  growing
• Current protection technologies have several
  limitations in supporting Federated Applications.
• Replication with CDP can ensure consistency for
  multiple applications across databases, hosts and
  storage arrays

29
QA / Feedback

• Please send any questions or comments on this
  presentation to EMC
• archer_gary_at_emc.com

Many thanks to the following individuals for
their contributions to this presentation. Rich
Walsworth Jason Iehl Peter Cutts Agnes Lamont
Puneesh Chaudhry