I have come across an example in a book I don’t quitefeel comfortable with. It concerns backing up a database by filegroup.
A database has three filegroups. One filegroup is backed up per night so that any given filegroup will be backed up once over a three day period. The transaction log is backed up daily.
The example states that if one of the file groups were to become corrupt during the day, the correct solution would be to restore the damaged filegroup only; and then apply the transaction log backup. The transaction log would of course be applied from the point just after the file group backup had been taken.
For this method to work, it must mean that when the transaction log is applied, any transactions that involve tables in the other filegroups will not be duplicated.
How does SQL Server manage this? Does a transaaction log restore actually rewind all transactions across all file groups from the restore start point - and then re-apply the transactions across all filegroups?
Dazed and confused
(N+, MCAD)
A database has three filegroups. One filegroup is backed up per night so that any given filegroup will be backed up once over a three day period. The transaction log is backed up daily.
The example states that if one of the file groups were to become corrupt during the day, the correct solution would be to restore the damaged filegroup only; and then apply the transaction log backup. The transaction log would of course be applied from the point just after the file group backup had been taken.
For this method to work, it must mean that when the transaction log is applied, any transactions that involve tables in the other filegroups will not be duplicated.
How does SQL Server manage this? Does a transaaction log restore actually rewind all transactions across all file groups from the restore start point - and then re-apply the transactions across all filegroups?
Dazed and confused
(N+, MCAD)