I know that if you stripe backups across multiple destinations, the backup process will assign a thread for each destination. I would like to verify how multithreading works in regards to filegroup backups. I've tried BOL, and google for
a while, and haven't found a clear answer. My question is really two-parts:
- If I back up multiple file groups, all to the same destination, does each filegroup backup get a thread of its own? For example: say i have filegroup
x and y. If I back up both filegroups to drive 1, will two threads be used?
- Could those filegroup backups be striped across multiple destinations to achieve even further multithreading? For example: say I have filegoup
x and y. If I stripe filegroup x's backup across drive 1 and 2, and I stripe filegroup y's backup across drive 3 and 4, will four threads be used?
Our full backups are reaching 9+ hours. The vast majority of the data in the database is varbinary(max) data. We are running SQl Server 2008 SP1 on Windows Server 2008 R2.
Any input, or links to good information about this subject will be greatly appreciated. Thanks!
View Complete Post