Performance Problems Question

a lot of our users get Sql Server timeout errors from what appear to be
pretty basic queries run from our web app. the server should be well
equipped hardware wise to deal with server load (dual proc xeon with nearly
4 gb ram raid 5). concurrent user connections average around 300; it handles
an average of 4 transactions a second. running the queries themselves via QA
in some cases takes an extremely long time. otehr times, they run in a
timely fashion.
I notice on perfmon a something that raised an eyebrow - the database is
spread over 2 filegroups, primary is about 4.5GB, secondary index filegroup
is about 2.3 GB. but we are getting very high disk queue values - averaging
50 over an hour or so.
While I know that as general rule of thum the threshold for disk queue is 2
* physical drives, 2 of the drives are hardly used. One drive contains only
backups, the second contains Diskeeper files, the 3rd is the OS drive. In
theory, only 2 of the drives are being written to during normal database
usage.
My question - is THIS normal (can I rule out the queue as a bottleneck),
given that we arent separating the filegroups over the RAID?I notice the system gets tweaked during Page Reads... probably indexes then?
"Ick" <nospamplease> wrote in message
news:uFZMAsphDHA.4088@.tk2msftngp13.phx.gbl...
> a lot of our users get Sql Server timeout errors from what appear to be
> pretty basic queries run from our web app. the server should be well
> equipped hardware wise to deal with server load (dual proc xeon with
nearly
> 4 gb ram raid 5). concurrent user connections average around 300; it
handles
> an average of 4 transactions a second. running the queries themselves via
QA
> in some cases takes an extremely long time. otehr times, they run in a
> timely fashion.
> I notice on perfmon a something that raised an eyebrow - the database is
> spread over 2 filegroups, primary is about 4.5GB, secondary index
filegroup
> is about 2.3 GB. but we are getting very high disk queue values -
averaging
> 50 over an hour or so.
> While I know that as general rule of thum the threshold for disk queue is
2
> * physical drives, 2 of the drives are hardly used. One drive contains
only
> backups, the second contains Diskeeper files, the 3rd is the OS drive. In
> theory, only 2 of the drives are being written to during normal database
> usage.
> My question - is THIS normal (can I rule out the queue as a bottleneck),
> given that we arent separating the filegroups over the RAID?
>
>