We are experiencing intermittent slowdows on our ASE 12.5.3 64-bit ESD #1
system and suspect a problem with checkpoint (both the checkpoint itself and
the transaction log purge).

It has been suugested that we increase housekeepeer free wite percent and
i/o batch size. I am also thinking about putting the logs on their own
segments and peerhaps a tempdb cache.

Has anyone had a similar problem? Does anyone have experience yo share on
these new parameters.

Sybase has suggested going to ESD#7 but that will take some time.

Thanks
Richard

Do you have any named caches other than "default data cache"?
What is your spinlock % in sp_sysmon? It is the first line under each cache
definition.

If you have spinlocks, then add a another named cache for a hot db like tempdb
or increase your "cache partitions" with sp_cacheconfig. We offen set cache
partitions at no less than 1/2 of our cpu engines.




On 1 Jun 2006 20:09:15 -0700,
in sybase.public.ase.performance+tuning
Richard Kinread <rkinread1 (AT) sympatico (DOT) ca> wrote:

Sybase TS is not responsible on suggesting ESD#7, I kind of piss off this
type answer from them.
Did Sybase TS provide CRs on this?

I would rather take a look at sysmon or using MDA for furthering
investigation.


"Richard Kinread" <rkinread1 (AT) sympatico (DOT) ca> wrote

TS suggested CR400907 and the description doesn't really match our problem.

I haven't been able to find anything in MDA that shows free checkpoints - do
you wheere to look?

When we get these slowdowns, our cpu utilization goes to 100% (i.e. all
engines at 100%). I suspect spinlocks but no spinlock activity shows up. I
am wondering if we are in a piece of code where Sybase doesn't measure what
is happening.

Thx
Richard Kinread

"Rey Wang" <reywang (AT) optonline (DOT) net> wrote

Post the sysmon output for 2 minutes sample with sp_helpcache,
sp_cacheconfig.



On Sat, 03 Jun 2006 12:52:51
-0700, Richard Kinread wrote: