Met a DBA from a company who are in business for ~2 years. During the conversation came to know they have never tested their backup file - not even once. Also, they have very rarely used DBCC CheckDB command. Should I call it as surprised or shocked?
It has become common to see Database Administrators to have fancy DB Backup plan, automate it and that's it. Their idea is whenever the need arises (read as disaster strikes) they can make use of it and restore the database. Although theoretically sounds like a good plan it actually isn't. Why?
All those efforts to take regular backups would become completely useless if those files can't be used to recover the database. One important question missed by many companies to whom I have consulted for - "Do we regularly make sure that we are able to restore a database from our backups?".
Despite backup process has succeeded how do we know it isn't corrupted (or) has some issues which don't allow it to be restored properly?
The bottom line is unless we do a database restore somewhere, we can't be completely sure that a given backup file is good.
At a high level, based on the business need we should first decide on:
- How much of "downtime" the system can have,
- How much of "data loss" is acceptable,
- After that put together a "Backup plan" to satisfy those requirements.
- Then the "Restore plan" should make sure all backup files are verified to be perfect and restorable in a test environment. Believe one would be already aware & using DBCC CheckDB often if not check the link provided at the end of this article.
DBA need to have a system that allows them to periodically review their plan and ensure they can get everything back up & running again.
Take away: I can't stress enough - Test your Backup file by periodically restoring in a test environment and make sure it's fine.
Recommended for further reading:
1. Paul Randal's The Accidental DBA (Day 8 of 30): Backups: Planning a Recovery Strategy
2. Award winning Maintenance script of Ola Hallengren
3. DBCC CheckDB FAQ by Kendra Little
It has become common to see Database Administrators to have fancy DB Backup plan, automate it and that's it. Their idea is whenever the need arises (read as disaster strikes) they can make use of it and restore the database. Although theoretically sounds like a good plan it actually isn't. Why?
All those efforts to take regular backups would become completely useless if those files can't be used to recover the database. One important question missed by many companies to whom I have consulted for - "Do we regularly make sure that we are able to restore a database from our backups?".
Despite backup process has succeeded how do we know it isn't corrupted (or) has some issues which don't allow it to be restored properly?
Our data is only as good as our last restorable backup
At a high level, based on the business need we should first decide on:
- How much of "downtime" the system can have,
- How much of "data loss" is acceptable,
- After that put together a "Backup plan" to satisfy those requirements.
- Then the "Restore plan" should make sure all backup files are verified to be perfect and restorable in a test environment. Believe one would be already aware & using DBCC CheckDB often if not check the link provided at the end of this article.
DBA need to have a system that allows them to periodically review their plan and ensure they can get everything back up & running again.
Take away: I can't stress enough - Test your Backup file by periodically restoring in a test environment and make sure it's fine.
Recommended for further reading:
1. Paul Randal's The Accidental DBA (Day 8 of 30): Backups: Planning a Recovery Strategy
2. Award winning Maintenance script of Ola Hallengren
3. DBCC CheckDB FAQ by Kendra Little
Comments