Unsolved
This post is more than 5 years old
27 Posts
0
1078
SQL database restore - what are least ssid(s) needed ?
Here is the situation that I am experiment:
- Two NetWorker backup server A and B, one SQL client C with 100 databases and one of the databases name called "xyz", and one SQL client D dedicated for restore purpose
- SQL server C databases backed up to server A, and the saveset specified as: MSSQL: (so it will backup all databases), full backup
now the database "xyz" need to restore to SQL server D from backup server B, assume all configurations are done properly, so mount the volume (cloned copy) to server B, and scan (user scanner -i -S ssids ...) only database "xyz" ssids to rebuild the client index for that particular database.
The restore client (SQL server D) seems not able to browse the database to restore, the thoughts on this is: if the saveset configured as MSSQL: it requires scan all ssids to rebuild index for browse even only database "xyz" need to be restored.
Any comment or lights to shed?
ble1
14.3K Posts
0
December 3rd, 2013 10:00
Do you see the database in index with nsrinfo?
ZENG1
27 Posts
0
December 3rd, 2013 11:00
I've found out the reason, with SQL module backup, each save set completed, it also generated a small incremental backup no matter what backup level setup was. For example, if the save set specified as "MSSQL:", after the full backup done, it also created this small incremental backup, and this saveset is needed to rebuild index for restore. So the minimual ssids required would be: the database save set ssids + the save set incremental ssid + the client index ssid.
Here are two questions to this incremental ssid:
- what is backing up on this incremental save set?
- does this logic also apply to other Microsoft applications? or non-Microsoft applications backup?
ble1
14.3K Posts
0
December 3rd, 2013 12:00
Incremental backup will give you transaction log backup. However, if you are referring to small ssid after DB backup, that is something else. Logic as such applies to NMM too if using VDI. With other modules it is different and it depends on how individual module is configured.
guenterH1
117 Posts
1
December 4th, 2013 00:00
This little save set always with level incremental appears after every kind of SQL backup done regardless database or log level backup.
This keeps some internal information about the database backup done.
It like a cover set, an ancor and you need this save set to be browsable too. So include in scanner command.
The key is (as far as I know) that NetWorker scopes on the backup when it starts but SQL server scopes on the time backup ends (key word: LSNs). This ancor save set is needed for recovery and browsing purpose.
Also as far as I know this behavior applies to MSSQL server backup via VDI interface only.
This does not apply to snapshot based SQL backup as done by NMM (nsrsnap_vss_save.exe).
From other applications/databases I cannot tell for sure but I've never seen such.
When using nsrinfo command to verify the client index please don't forget to browse the mssql name space (nsrinfo -n mssql .......).
ZENG1
27 Posts
0
December 6th, 2013 07:00
It makes sense, the snapshot backup will restore the time to take snap and leave the database consistency to VSS handle and VDI will have to depends on SQL mechanism on LSNs for consistency.