Microsoft Exchange 2000 Operations Guide — Version 1.080
Disaster Recovery Procedures
Chapter 2, “Capacity and Availability Management,” examined ways of minimizing
system failures. As mentioned there, to reduce overall downtime you need to look at how
frequently a system is down, alongside how long it takes to bring a system back up again.
For more information on Availability Management, refer to Chapter 2.
It makes no sense to have a sound, well-exercised backup strategy unless it’s matched with
a similarly mature recovery strategy. Backups are meaningless if you can’t restore service
using them.
Exactly how you perform disaster recovery will depend on how much money you are
willing to spend alongside which backup products you want to use. Third parties offer a
variety of solutions, including mailbox level backup, and even message level backup in
some cases. Another alternative is real-time byte level replication. You will find a list of
third party vendors offering backup solutions for Exchange 2000 at the following Web
site:
http://www.microsoft.com/exchange/thirdparty/E2Ksolutions.htm#backup
Whichever tools you use, the Operations Manager will need to ensure that the disaster
recovery procedures meet the following criteria:
Backup is performed regularly and reliably.
Your data is protected against fire/theft/natural disaster.
Recovery can be performed reliably and quickly.
You regularly do test restorations on an offline server to ensure that backups are being
correctly created.
You regularly run disaster recovery drills to keep skill levels high and ensure that your
recovery procedures are up to date.

Backing Up

The rest of this section assumes that you are using Windows NT Backup as your backup
and restore software. However, much of the information contained here is relevant which-
ever backup solution you choose to adopt.
One of the major considerations when performing a backup is how long it will take. When
performing an online backup of stores on your servers running Exchange, you suspend
other online maintenance activities. Therefore, to allow appropriate time for the other
online maintenance activities to occur each night, you need to minimize the length of time
backup takes. Also, if it takes a long time to back up these servers, it will almost certainly
take a long time to restore them, and to meet your SLAs you will want to keep your
recovery times to a minimum.