High cpu utilization on sqlserv during backupcopies, slow ..

[Peejay62]

Hi,

I noticed during copy process a very high cpu utilization for sqlserver on the backupserver. If you look at a copyjob, everytime it skips to the next vm to process, the sql server tops at 100% and stays there for quite a while. It also takes quite a while to start processing the next VM. Even when only one backupcopyjob runs it shows this behaviour. When I run some copies next to one another, the system almost gets unresponsive. You see it in the graph of the copyjob, there is some throughput, than for quite a while nothing (in this period the sql is @100%) and then again some activity and so on. The "building VMlist" step as the copies start also takes a long time, from 4 minutes to even almost an hour...
I run 2 backupservers that are almost identical by design and workload. One is just working fine, even with parallel copyjobs, it doesn't show this behaviour at all. Sometimes the sqlserver gets a spike, but that's it. I have the overall feeling something is wrong on this backupserver, i cannot pinpoint it though. I cannot imagine that, given the information I have on the way the sql environment is used (mainly storing metadata about jobs and backuppoints and so on) this is normal behaviour. There is still an express version on it, but with appx 200 vms processed over 20 jobs, I don't consider this having reached the limits of the database.
The backupserver is a Quad core 2.3 GHZ Amd Opteron 6176SE, 8 GB mem with windows 2008 R2 SP1.

Any input would be appreciated,

Bye, Peter

[chrisdearden]

sounds like a good time for Patch 3 to be released , which should help with a few of those issues. Did the other backup server have a full version of SQl on it ?

[Peejay62]

Hi Chris,

No, the other server also has express installed. I am experiences this behaviour since a couple of days. I added one new job last weekend. That's the only thing that changed, up to last week everything ran fine. maybe I reached a limit now?
The other server has even more load though. When i take a look at the size of the database (size of the files in the sql directory) the db on the "good" server is even bigger.

[chrisdearden]

With Express it can be as large as 10GB. I wonder if there is a bit of SQl maintenance that coudl be done - might be worth logging a ticket just in case ( that said I would imagine patch 3 will be the first recommendation )

[foggy]

The patch is already available, updating would definitely be the first troubleshooting step as reduced database load is among introduced improvements.

[Peejay62]

Ok, it comes at a good moment. Unfortunately I am not allowed to throw in a patch at any given moment, so as soon as as possible, I will update, hopefully it solves the issue.

thanks, Peter