The system log is critical in the event of a system failure or performance degradation.
You can use the following information
to determine the sequence of events leading up to the failure and to aid recovery:
* Observations
An important part of the DBA's job
is to notice changes in the system and to anticipate problems.
Observations of unusual activity should be noted.
Even a note as simple as "The system seems sluggish"
might hold valuable clues in the event of a subsequent system failure.
* System changes
The DBA should record all changes made to the hardware,
the operating system, and the database system.
Entries should be in chronological order and should be complete but without unnecessary details.
* System failures
Anytime a disk drive or other component fails,
the event should be documented in the system log.
This information can be valuable in determining trends of component failures.
* Backup and restore operations
It is not necessary to update the log every time the system is backed up. However, requests for data to be restored should be logged
to show patterns in user behavior as well as trouble spots in an application or a database schema.
* Scheduled maintenance
When scheduled maintenance is performed,
the DBA should make a note of what was done to the system.
This information can be the starting point of investigating a system failure that occurs soon after a scheduled maintenance.
By keeping track of critical events as well as of configuration information,
you can determine where a problem has occurred
and know how to get the system back to where it should be.
Source:
Microsoft SQL Server 2000 Administrator's Companion eBook