Re: longops vs sql

DA Morgan wrote:
> Ben wrote:
>
> > Sorry, that is assuming that 300 disk I/Os can be performed per second.
> > That is just an estimate.
>
> Bad assumption. Are you familiar with running
> DBMS_STATS.GATHER_SYSTEM_STATS?
>
> Morgan's Library
> www.psoug.org
> click on SYSTEM STATISTICS
>
> This will give you accurate information. For my IBM ThinkPad:
>
> CPUSPEEDNW 1364.342
> IOSEEKTIM 13.923
> IOTFRSPEED 4096
> SREADTIM 5.714
> MREADTIM 5.903
> CPUSPEED 1378
> MBRC 13
>
> Not a 300 in sight.
> --
> Daniel Morgan
> University of Washington
> Puget Sound Oracle Users Group

based on your figures, isn't that showing that 300 I/Os would take 1.77 secs? I would think that a ThinkPad could be a little slower than a more robust machine. If I'm calculating that wrong please show me what I'm missing.

MREADTIM is close to SREADTIM but I used MREADTIM. This is telling me that it takes 5.903 millisec for a multi block read request. If I perform 300 multiblock read requests (300 I/O) that should take 1770.9 millisecs. Convert that to seconds and it's 1.7709 secs/ 300 I/O. Which is longer than my original assumption but that is also on a laptop and not by more than a second.

Am I close to correct in my calculation there?

Thanks Received on Thu Sep 07 2006 - 14:36:06 CDT