Disk I/O

I would like to confirm what Perfmon counters I should be looking at to esta
blish if the system running our Production Instance of SQL Server, has an I/
O bottleneck. I have been focusing on Current Disk Queue Length counter whic
h during peak times (a 2.5
hour interval) peaks around 200+ for extended intervals. We have storage on
a SAN (EMC disk array with RAID 10). The databases reside (both data and tra
nsaction logs) on a single logical drive that comprises of 4 physical disks
on the EMC. Current queue l
ength of 200+ (for extended periods of time) seems like a problem, but Micro
soft expert says that we should be looking at Avg Queue Length counter. I ne
ed to know what counters can help us determine if in fact there is I/O bottl
eneck
Thanks, ShehlaI agree that looking at average disk queue might give a better overall
picture but if your getting current disk queues of 200+ for extended periods
of time that will be an issue. You should also check with who ever
maintains the SAN to see what tools they have for monitoring the disk
activity.
Andrew J. Kelly SQL MVP
"Shehla Arshad" <anonymous@.discussions.microsoft.com> wrote in message
news:D46831D9-D706-4944-ADC4-FEBCBE13CFE8@.microsoft.com...
> I would like to confirm what Perfmon counters I should be looking at to
establish if the system running our Production Instance of SQL Server, has
an I/O bottleneck. I have been focusing on Current Disk Queue Length counter
which during peak times (a 2.5 hour interval) peaks around 200+ for extended
intervals. We have storage on a SAN (EMC disk array with RAID 10). The
databases reside (both data and transaction logs) on a single logical drive
that comprises of 4 physical disks on the EMC. Current queue length of 200+
(for extended periods of time) seems like a problem, but Microsoft expert
says that we should be looking at Avg Queue Length counter. I need to know
what counters can help us determine if in fact there is I/O bottleneck
> Thanks, Shehla|||In addition to Andrew's comments... average disk q is better, and average
queue > 2 is bad... However the number that comes from perfmon is for the
drive letter... you must divide the perfmon number (200) by the number of
physical drives behind the letter, (4), so your average queueing is still
50( way too high.)
Wayne Snyder, MCDBA, SQL Server MVP
Computer Education Services Corporation (CESC), Charlotte, NC
www.computeredservices.com
(Please respond only to the newsgroups.)
I support the Professional Association of SQL Server (PASS) and it's
community of SQL Server professionals.
www.sqlpass.org
"Shehla Arshad" <anonymous@.discussions.microsoft.com> wrote in message
news:D46831D9-D706-4944-ADC4-FEBCBE13CFE8@.microsoft.com...
> I would like to confirm what Perfmon counters I should be looking at to
establish if the system running our Production Instance of SQL Server, has
an I/O bottleneck. I have been focusing on Current Disk Queue Length counter
which during peak times (a 2.5 hour interval) peaks around 200+ for extended
intervals. We have storage on a SAN (EMC disk array with RAID 10). The
databases reside (both data and transaction logs) on a single logical drive
that comprises of 4 physical disks on the EMC. Current queue length of 200+
(for extended periods of time) seems like a problem, but Microsoft expert
says that we should be looking at Avg Queue Length counter. I need to know
what counters can help us determine if in fact there is I/O bottleneck
> Thanks, Shehla