Comments on: What Happened Here?

By: Michal Poreba

Michal Poreba — Sun, 25 Mar 2012 13:47:00 +0000

To me it looks like the SQL Server cannot meet the required recovery interval. I would like to check few things

– recovery interval settings

– free space on the disk used for log

– any potential problems with I/O on that disk

– size and fullness of log file, settings of log file

I don’t think a code or small recovery interval is to blame. You cannot really answer performance questions with ‘you are doing too many updates’. Transaction logs do not cause checkpoints so frequency will not change anything. I don’t think there are too many scenarios when one would actually want to change recovery interval from the default 60 seconds, but I would check it just to see what it is.

What I think it is, is the log file may be too small (space on the storage device or size limit). The log file access may be too slow. That is were I would start looking for an answer.

By: ThomasLaRock

ThomasLaRock — Sat, 24 Mar 2012 02:32:00 +0000

In reply to Shaunjstu. Shaunjstu, Would a transaction log backup have anything to do with what you see in those charts?

By: Shaunjstu

Shaunjstu — Thu, 22 Mar 2012 19:50:00 +0000

I agree with Tracy. Looks like a high volume server with some code that issues checkpoints frequently, or as Tracy mentioned, a small recovery interval configuration on the server. What action should a DBA take here? I don’t think any is needed. (Well, maybe increase the frequency of t-log backups if this is in fact caused by data changing frequently in a db in full recovery mode.)

By the way, I love this little series! This is a great idea!

By: Tracy McKibben

Tracy McKibben — Thu, 22 Mar 2012 18:36:00 +0000

Looks like a system with plenty of memory, configured with a low recovery interval….