On 1/14/2010 12:07 PM, Greg Smith wrote:
Andy Colson wrote:
On 1/13/2010 11:36 PM, Craig Ringer wrote:
Yes. My 3ware 8500-8 on a Debian Sarge box was so awful that launching a
terminal would go from a 1/4 second operation to a 5 minute operation
under heavy write load by one writer. I landed up having to modify the
driver to partially mitigate the issue, but a single user on the
terminal server performing any sort of heavy writing would still
absolutely nuke performance.
On a side note, on linux, would using the deadline scheduler resolve
that?
I've never seen the deadline scheduler resolve anything. If you're out
of I/O capacity and that's blocking other work, performance is dominated
by the policies of the underlying controller/device caches. Think about
it a minute: disks nowadays can easily have 32MB of buffer in them,
right? And random read/write operations are lucky to clear 2MB/s on
cheap drivers. So once the drive is filled with requests, you can easily
sit there for ten seconds before the scheduler even has any input on
resolving the situation. That's even more true if you've got a larger
controller cache in the mix.
That makes sense. So if there is very little io, or if there is way way
too much, then the scheduler really doesn't matter. So there is a slim
middle ground where the io is within a small percent of the HD capacity
where the scheduler might make a difference?
-Andy
--
Sent via pgsql-performance mailing list (pgsql-performance@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance
