July 19, 2006 at 6:08 am
Hi everyone,
I was monitoring the disk time on my SQL box.
It has a 2 Terabytes capacity and its on a SAN. Is it true that the disk time shodlnt exceed 20, or this could mean there is a bottleneck somewhere.
My disktime goes from 0 - 1500 - 6000 - 4549 - 0 again.
do I have a bottleneck ? and also should any considerations be taken for SANs ?
What kind of maintainance tasks can i do to improve the SAN ?
Also, is it possible to save perfmon counters in a txt file or DB table, so that I can accurately analyse it ?
July 19, 2006 at 6:31 am
Hi John!
We have faced that kind of problem once and we fixed it with hardware vendor. I think it's not a bottleneck but something related with your disk controllers. Apply your vendor.
July 19, 2006 at 10:57 pm
It could mean an I/O bottleneck, or it could mean that there are some very inefficient operations running against the database server. Usually it will be the latter. Since your disk utilization seems to be spiking, this seems likely. Huge scans (table or index), backups, dbcc checkdbs can cause heavy disk utilization for brief (and sometimes not-so-brief) periods. Run a profiler trace at the same time to see if you log any queries with high read values. Set the trace filter to capture moderate duration queries, say 200ms or so.
As for logging perfmon counters, there are a few ways.
From the perfmon console, expand "Performance Logs and Alerts" and clisk "Counter Logs". Create a new log, add your counters, and on the "Log Files" tab, change the "Log file type" to CSV or TSV.
Another way, if your running XP or 2003, is to use the typeperf.exe console utility. typeperf /? should tell you all you need to know.
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply