Enterprise Data Replication

1
Enterprise Data Replication

• By Bob Bryar
• March 2002 for SearchStorage

2
Objective

• Using state of the art technology, IT Executives
  must execute tactically against a strategic
  vision that will provide all Servers within the
  corporation, a seamless connection topology to
  their most valued corporate resource, data
  storage
• The chosen topology needs to meet several key
  criteria
• Return On Investment (ROI)
• Increased Operational Efficiency
• Maintain a High Level of End User Satisfaction
• Meet Business Drivers for Recovery

3
The Data Struggle
The Data Explosion

• Managing systems diversity and dispersion of data
• Ensuring the speed and availability of data
• Ensuring the reliability and availability of
  systems
• Ensuring Recovery of Data is obtainable

The Challenge
Which tools will you utilize to manage and
protect the data?
4
Companies are seeking Technological Solutions to
many Storage issues!
1
Source Gartner Dataquest, September 2000
5
Evolution of Storage Architectures
Network Centric
10/100, GbE, ISCSI
6
Storage Spending Will Surpass Servers in IT
Storage is now a separate, strategic decision
Server versus Storage Spending Research from 1999
Server versus Storage Spending projections for
2003
Storage Spending
Server Spending
By 2003, servers will become peripheral to
storage Forrester Research
7
Processor / Server Complex
Application
O/S File System
DB I/Os
Overview of Typical SCSI I/O Command to Disk
Logical Volume Manager
System Raw Partition
Substitute SCSI HBA For FC HBA In SAN Environment
SCSI Device Driver
SCSI I/O INTERNAL
SCSI HBA
SAN
Block I/O
SCSI I/O EXTERNAL SAN
8
Data is overwhelming the IT department

• Virtual Storage
• Storage Area Networks
• Availability Strategies
• Replication
• Mirroring
• Infrastructure
• DWDM
• SAN
• Monitoring

How do you handle the growth?
Storage Resources
Security
Archive
Replication
Performance
Database Server
Capacity Allocation
Backup
Asset Management
Tape Library
Disk Arrays
9
Change Management

• Configure hardware
• CWS build
• Install nodes 72 hrs.
• Configure nodes to
• ADSM server 3 hrs.
• Application Restores

72 hrs. to 3 hrs.
J40
J40
R50
Thin Node High Node Wide Node
J40
Operating System Kernel1 Kernel2 Kernel3 Patches U
pgrades
Platform OS
Application Layer
Common Method
Kernel
R50
Thin Node High Node Wide Node
J40
10
Elimination of Backup Window
STORAGE

• Off-host backup and recovery
• Application-aware
• Direct disk to tape I/O via external data mover
  
• Enhanced performance
• Multi-platform and scalable

WEB
mobile
FIREWALL
11
Availability Must be Engineered into the Data Flow
D1
D2
D3
D4
SAN Switch
Router Firewall
Application Server
Static Web Servers
Static Web Servers
Database Servers
Cache Server
Application Server
Static Web Servers
Static Web Servers
LB Server
Application Server
12
Integration must occur between all Vendors

• xSP market is expanding and getting legitimate

• Outsourcing trend is not slowing down
• MUST create service levels with teeth

• All moving to fully managed solutions

Router Firewall
Application Server
Static Web Servers
Static Web Servers
Database Servers
Cache Server
Application Server
Static Web Servers
Static Web Servers
LB Server
Application Server
13
Common operating platforms
14
The Data Management Struggle

• Gartner estimates that by the end of 2001, 25 of
  corporate IT organizations will outsource or
  consider outsourcing some aspect of storage
  infrastructure or management

15
Discussions about internal recovery

• Why cant we do it ourselves?
• We feel like were paying a lot of money and not
  getting value!
• Ive got all the components
• Network
• Diverse Centers
• Knowledge in High Availability
• Lower Cost of Technology

• What are companies struggling with?
• Change Management
• Cost of maintaining dual environments
• Validating the strategy and capability
• Methodology
• Effort

16
What is the data protection trend?
Traditional Hot Site
Advanced Recovery Solutions
Operational Availability on a Continuous Basis
Network Storage
Customer Data Center
Tomorrow
17
Choosing the Technology

• IBM, STK (PPRC)
• EMC (SRDF)
• HDS (TrueCopy)
• STK (PPRC)
• Network Appliance (SnapMirror)

18
Three - Node ring Using DWDM
OPTera Metro
ESCON
ESCON
FICON
FICON
. .
. .
GEOPLEX
GEOPLEX
Fibre Channel
Fibre Channel
Gigabit Ethernet
Gigabit Ethernet
OPTera Metro
OC-3/12/48
OC-3/12/48
Test Center
Production Center
. .
GbE
ESCON
GEOPLEX
Recovery Center
OC-3/12/48
FICON
FC

• All three sites appear as one processor complex

19
Storage with Recovery Objectives
EMC
MVS
MVS
NT
Storage
NT
Dark Fiber / Channel Extension
Remote Site
Local Site
NT
Storage
Tape
Tape
20
Replication from Production to Recovery Location
Customer Production 1
TSC - Carlstadt
200GB
T1 / T3
F760
SAN
400 GB
F760
SAN
Customer Production 2
Customer Shared Storage
200GB
F760
SAN
T1 / T3
F760
Security via O/S Tools
21
Data Replication UtilizingSNAP Technology
Customer Site
Recovery Center
Switch

• Switch

• ASYNCHRONOUS IP Data Replication
• With Restore Capability

PROCESSORS
Managed Communications Links T1 T3 OCX
SNAP Technology Data Replication
MIPs / MHz Recovery Shared / Dedicated
22
Mainframe Data Replication Services using IP
Customer DATA CENTER
Recovery Center
Switch

• Switch

• ASYNCHRONOUS IP Data Replication
• With BCV Restore Capability

IBM OS/390 PROCESSOR / UNISYS 2200
Comdisco Managed Communications Links T1 T3
OCX

• MIPs / MHz Recovery
• Future Implementation

• Symmetrix SRDF
• Data Replication

23
Customer Provided Storage Services Many Systems,
Merging Technologies
Midrange Systems
Mainframe
Switched Fabric
SRDF / Fibre Channel
Continuity Service Provider
MVS Storage