Re: V$FILESTAT Average Write Times

On 20 Sep 2001 07:02:27 -0700, mark.powell_at_eds.com (Mark D Powell) wrote:

>
>The average IO time of v$filestat can be distorted by a bad IO so the
>numbers are not necessarily reliable. You need to compare the Oracle
>stats to device stats obtained at the OS level and eliminate any
>periods where hung IO's and other problems existed as opposed to just
>load related problems.
>

That is my experience too with EMCs. The V$FILESTAT info is not reliable in that environment. We need to do it at OS level. And even then it is not 100% reliable: when the cache saturates and the system becomes truly IO bound, then we experience wild variations in measured times. Averages will then of course be highly skewed. In the EMC-based system I'm most familiar with, this happens at around 0.5Gb of continuous writes, which is also the cache apportioning for the individual LUs.

I'm told there are specific tools that will measure IO times accurately for this type of box, but I've never seen them. But I'm happy with the info I get from good old "sar -d": more interested in relative IO balance between LUs than the actual IO times I get on each one. Although I have an abnormal situation in one of them where the O times are consistently much higher than with the others. Still haven't found an explanation.

Cheers
Nuno Souto
nsouto_at_optushome.com.au.nospam Received on Thu Sep 20 2001 - 09:25:51 CDT