NSI Software Product Overview 11182004

1
NSI Software Product Overview11/18/2004

• .
• David J. Demlow
• Vice President, Product Management
• NSI SOFTWARE
• ddemlow_at_nsisoftware.com

Warning No part of this document may be
reproduced or transmitted in any form or by any
means, electronic, or mechanical, for any reason,
without the express written permission of NSI
Software. The information in this document is
subject to change without notice. Companies,
names and data used in examples herein are
hypothetical and/or fictitious unless otherwise
stated. Note Product names mentioned herein
may be trademarks and/or registered trademarks of
their respective companies.
2
Agenda

• Introduction to NSI Software
• .
• The Technology behind Double-Take
• Double-Take Business Solutions
• GeoCluster Overview
• Q A

3
So, here is your Windows Network
4
NSI Software Overview

• Leading Provider of Data Replication and High
  Availability Software and Services
• Founded in 1991
• Over 200 employees in three offices Hoboken, NJ,
  Indianapolis, IN and Southboro, MA
• Award-winning products Double-Take and
  GeoCluster
• Over 20 patented technologies
• Comprehensive Professional Services Offerings
• Prestigious Microsoft Certifications

5
Continuous Data Replication for Windows

• Data Protection that is
• On Time
• Continuous Protection
• RPOseconds / minutes
• On Line
• Immediate Availability
• RTOminutes
• Off Site
• Unlimited Distance
• Minimum bandwidth

6
Award Winning Solutions

• Real-time data protection
• Automatic server failover
• Disaster recovery - LAN and WAN environments
• Windows NT4, 2000, WPNAS and 2003
• (Server, Enterprise / Advanced Server, and
  DataCenter)

• Provides redundancy of MSCS storage
• Allows MSCS cluster to be separated over IP
• Based on DoubleTake replication engine
• Windows NT 4, 2000, WPNAS and 2003
• (Enterprise / Advanced Server / DataCenter)

7
How Double-Take Works

• Select Source Files and Target Path
• Double-Take compares source and target files and
  then synchronizes the differences at a block
  level.
• On-going changes to selected files are captured
  for replication as they occur.
• Changes are transmitted continuously or per
  user-defined conditions, such as bandwidth limits
  or start/stop conditions.
• Target receives and applies the changes to target
  volumes.
• Target server monitors the source server and can
  initiate failover automatically or manually.
• Failback releases the identity of the source
  allowing original source to rejoin the network.
• The restore function returns current data back to
  original machine.    

8
Create a new Replication Set
9
For a File Server, this might be \APPS \DATA
For a Web Server, \INETPUB
For a SQL Server, \DATA \LOGS
For an Exchange Server, \MDBDATA \MTADATA
10
Drag the Repset over a Target server and a
connection is started
11
Configure server failover options (bottom)
12
Then watch the mirror complete!
13
Green Everything is good
14
At zero (OPTIONAL) manual safety on failover
15
One-button Fail-Back
16
(No Transcript)
17
Replication Options

• Where does it operate?
• Software or Hardware
• In the Host, Controller, or Storage
• What does it capture and replicate?
• File Byte Changes
• Disk Block Changes or SCSI I/O
• How is performance impacted
• Periodic or Continuous?
• Synchronous or Asynchronous?
• Is it in the I/O path?
• Is I/O response delayed
• How are in flight transactions handled
• How is data transmitted?
• Network (IP)
• Fibre Channel Fabric
• Protocol Efficiency
• Is it available online or is it offline?

18
File Change Replication
Application
Application
File Change 487 Path File Op Start Length DATA
. . . . . . . . . .
Operating System
Operating System
487
. .
File System
File System
328
328
327
327
Hardware Layer
Hardware Layer
326
326
19
Synchronous vs. Asynchronous
20
Asynchronous Wins in Performance
21
Double-Take Advantages

• Software Based, File Centric, Byte Level
  Replication
• Server/Storage Independent
• Include or exclude files, directories, wildcards
• Byte level file system changes captured for
  replication
• Minimum amount of data transmitted
• STAR Technology
• Sequential Transfer
• Asynchronous Replication
• Negligible impact on server, application and I/O
  performance
• Live File System on Target
• No Volume Mount / Check
• Read Only Usage such as Backup, Replication
  Chaining

Tape Backup Hours / Days
Synchronous Mirror
Clustering
Seconds / Minutes
Recovery Time (RTO)
Five 9s
Recovery Point (RPO)
22
Business Solutions
23
HIGH AVAILABILITY

• For the purposes of this presentation,
• High Availability is defined as a method by
  which user data and/or applications are protected
  and continue to be available to the user
  community in order to allow the user community to
  remain productive.
• This level of survivability assumes and
  requires that the remainder of the computing
  environment is functional meaning that the
  users workstations continue to have power and
  connectivity to the server resources and the
  network topology has not been significantly
  altered

24
High Availability
Production (source) Servers
FS1 DB
10.9.9.1
WWW System Services started
FS2
10.9.9.2
dB System Services started
Redundant (target) Server
FS3 WWW
10.9.9.3
FSDT
10.9.9.252
FS4
10.9.9.4
FS5 F/P
10.9.9.5
25
DISASTER RECOVERY

• For the purposes of this presentation,
• Disaster Recovery is defined as a method by which
  network information is protected and continues to
  be available in the event that the computing
  environment is critically impacted.
• This level of survivability makes no assumptions
  and/or requirements for other surviving
  components.

26
Disaster Recovery
Disaster Recovery
Off-Site Storage
27
BACKUP RESTORE

• For the purposes of this presentation,
• There is no such thing as Backup
• The idea of backing up
• or writing data to tape/optical is simply the
• PREPARATION FOR RESTORE
• it is also that series of tasks that make
  auditors happy.

28
Backup RecoveryWhy not backup at 3PM ?

• Open Files on all servers
• high CPU on backup server
• high network during backup
• I/O results in system-crash

Tape Backup
29
Backup RecoveryWhy not backup at 3PM ?
FS1
Seattle

• Open Files on all servers
• high CPU on backup server
• high network during backup
• I/O results in system-crash
• Every office backing up itself (managed by
  non-I/S personnel)

FS2
Dallas
FS3
Louisville
FS4
Nashville
FS5
Lancaster
Tape Backup

• Off-Site Courier Services

30
Enhanced Backup

• no Open Files - without agents
• no CPU issues
• no network bandwidth limits
• no stability concerns
• truly Centralized Backup

Off-Site Storage
Tape Backup
31
Combining Replication and VSS Snapshots
Double-Take Replication
REMOTE SITE running Windows 2003
DATA CENTER running Windows 2003
32
and SANs

• SAN to SAN
• Remote Data Protection
• Consolidation of storage onto SAN
• SAN as replication transport
• Single Server (S to T)
• IP over Fibre Channel

33
Cluster Disaster Recovery
MSCS Cluster Nodes
Double-Take
Double-Take
Replication from Node 1
Replication from Node 2
34
Cluster Disaster Recovery
MSCS Cluster Nodes
Double-Take
Double-Take
Replication from Node 1
Replication from Node 2
35
and any time that you need multiple copies of
your active data
36
MSCS Limitations

• Specific Hardware requirements
• Single copy of data, failure points
• Geographic Limitations

• The Availability of a Cluster with the Protection
  of Real-Time Data Replication

37

• Leverages patented Double-Take replication engine
• Adds data redundancy to MSCS by providing
  replicated volumes
• Tight Integration with MSCS
• Supports MSCS aware applications
• SQL Server
• 2000 and 7.0
• Exchange Server
• 2003 and 2000
• IIS
• Others
• Now Certified for Windows Server 2003 and 2000!

38
http//www.NSISOFTWARE.com
39
Background Slides How does it REALLY work ?

40
How Replication really works from an EXCHANGE
perspective
41
How Replication really works from an EXCHANGE
perspective

• Client sends message

• Mail client sends to Exchange server

42
How Replication really works from an EXCHANGE
perspective
IMMEDIATE (to DISK) for LOG dB / Recipient
/ MessageTo/Cc/Opt/XxXxXx

• Client sends message

TO DATA-STORE in MEMORY dB / Recipient
/ MessageTo/Cc/Opt/XxXxXx

• Mail client sends to Exchange server

• Server updates LOG and DATA

43
How Replication really works from an EXCHANGE
perspective
IMMEDIATE (to DISK) for LOG dB / Recipient
/ MessageTo/Cc/Opt/XxXxXx

• Client sends message

TO DATA-STORE in MEMORY dB / Recipient
/ MessageTo/Cc/Opt/XxXxXx

• Mail client sends to Exchange server

(DATA PAGES eventually paged to disk)

• Server updates LOG and DATA

• Server OS generates

TWO DISK WRITES
one to LOG.CHK one to PRIV.EDB
44
How Replication really works from an
OPERATING SYSTEM perspective
IMMEDIATE (to DISK) for LOG dB / Recipient
/ MessageTo/Cc/Opt/XxXxXx
Disk Operation 326 Path d\EXCHSRVR\MDBDATA\LOG
S File LOGx.CHK Operation Write Start
1720 bytes Length 42 bytes Data
To/Cc/Opt/XxXxXx
TO DATA-STORE in MEMORY dB / Recipient
/ MessageTo/Cc/Opt/XxXxXx
(DATA PAGES eventually paged to disk)
Disk Operation 487 Path d\EXCHSRVR\MDBDATA\ Fi
le PRIV.EDB Operation Write Start 92324
bytes Length 42 bytes Data To/Cc/Opt/XxXxXx

• Server OS generates

TWO DISK WRITES
one to LOG.CHK one to PRIV.EDB
45
How Replication really works from an
OPERATING SYSTEM perspective
Disk Operation 326 Path d\EXCHSRVR\MDBDATA\LOG
S File LOGx.CHK Operation Write Start
1720 bytes Length 42 bytes Data
To/Cc/Opt/XxXxXx
Application
Operating System
File System
Hardware Layer
46
How Replication really works from an
DOUBLE-TAKE perspective
Disk Operation 326 Path d\EXCHSRVR\MDBDATA\LOG
S File LOGx.CHK Operation Write Start
1720 bytes Length 42 bytes Data
To/Cc/Opt/XxXxXx
Application
Disk Op 326 Path File Op Start Length DATA
Operating System
File System

• DBLHOOK (driver) asks
• Is the Source connected to a Target ?
• YES

Hardware Layer

• Is it a Write or Read or ???
• WRITEYES

• Is the file or directory supposed to
• be replicated? YES In the Rep Set

47
How Replication really works from an
DOUBLE-TAKE perspective
Application
Disk Op 326 Path File Op Start Length DATA
Disk 327 Path File Op Start Length DATA
Disk Op 487 Path File Op Start Length DATA
Disk 328 Path File Op Start Length DATA
Operating System
. . .
487
. .
File System
328
327
Hardware Layer
326
48
How Replication really works from an
DOUBLE-TAKE perspective
Application
Application
Disk Op 487 Path File Op Start Length DATA
. . . . . . . . . .
Operating System
Operating System
487
. .
File System
File System
328
328
327
327
Hardware Layer
Hardware Layer
326
326
49
How Replication really works from an
DOUBLE-TAKE perspective
Its the same data in the same order
Application
487
. . . . . . . . . .
Operating System
File System
Disk Operation 326 Path d\EXCHSRVR\MDBDATA\LOG
S File LOGx.CHK Operation Write Start
1720 bytes Length 42 bytes Data
To/Cc/Opt/XxXxXx
328
T\FS1-D\EXCHSRVR\LOGS\
327
Hardware Layer
326
50
Replication Summary
Writes Only - not Reads Data selectable at a
File or Directory Level Transaction-based
Replication
Applications
Applications
Operating System
Operating System
File System
File System
Disk Hardware
Disk Hardware