High CPU during transaction log backups!

High CPU during transaction log backups!
I'm having issues with a SQL server where during
transaction log backups, the CPU usage us on average 20%-
25% higher than during normal. We have timed this out
using performance monitor and know for sure it's the
transaction logs at this point. If we turn off transaction
logs, the CPUs continue normal without the 20%-25% spike.
We have our data volume physically separate from our log
volume, and our backup volume again separate from both of
these. The log volume and the backup volume also are on
separate scsi channel.
While watching the performance monitor, the physical drive
stats are flat line with no queued requests on any of the
volumes so it does not appear to be disk io issue. All
raid is HW raid on a controller with 128 meg of cache,
which according to its stats, the controller is barely
breathing hard.
When watching the sysprocess, the last wait type is always
ASYNC_DISKPOOL_LOCK, which I cannot find much info on
other than it occurs during backups, and look at disk io
issues, which again, don't appear to be the issue.
I'm at a total loss. I have 8 other SQL servers with
similar configs that do not show this behavior. All boxes
are configed the same, meaning HW, software an settings.
Any other ideas?!?!
Configuration:
HP Proliant
Quad 700 Xeons
2.5 gig ram
Data drive 4 * 72 gig 10k scsi drives raid 5
Log drive 2 * 72 gig 10k scsi drives raid 1
Backup drive 3 * 72 gig 10k scsi drives raid 5
Win2000 SP2
SQL2K SP3a
-RossRoss
Is it sill under 80%-90% ? If it is, don't worry.
"Ross" <RMNornesNOSPAM@.NOSPAMjtmd.com> wrote in message
news:1e2d101c45578$fa214f60$a601280a@.phx
.gbl...
> High CPU during transaction log backups!
> I'm having issues with a SQL server where during
> transaction log backups, the CPU usage us on average 20%-
> 25% higher than during normal. We have timed this out
> using performance monitor and know for sure it's the
> transaction logs at this point. If we turn off transaction
> logs, the CPUs continue normal without the 20%-25% spike.
> We have our data volume physically separate from our log
> volume, and our backup volume again separate from both of
> these. The log volume and the backup volume also are on
> separate scsi channel.
> While watching the performance monitor, the physical drive
> stats are flat line with no queued requests on any of the
> volumes so it does not appear to be disk io issue. All
> raid is HW raid on a controller with 128 meg of cache,
> which according to its stats, the controller is barely
> breathing hard.
> When watching the sysprocess, the last wait type is always
> ASYNC_DISKPOOL_LOCK, which I cannot find much info on
> other than it occurs during backups, and look at disk io
> issues, which again, don't appear to be the issue.
> I'm at a total loss. I have 8 other SQL servers with
> similar configs that do not show this behavior. All boxes
> are configed the same, meaning HW, software an settings.
> Any other ideas?!?!
> Configuration:
> HP Proliant
> Quad 700 Xeons
> 2.5 gig ram
> Data drive 4 * 72 gig 10k scsi drives raid 5
> Log drive 2 * 72 gig 10k scsi drives raid 1
> Backup drive 3 * 72 gig 10k scsi drives raid 5
> Win2000 SP2
> SQL2K SP3a
> -Ross