Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: "Reasonable" Wait statistics??

Re: "Reasonable" Wait statistics??

From: Ricky Sanchez <rsanchez_at_more.net>
Date: Fri, 14 Jun 2002 02:21:19 GMT
Message-ID: <3D095381.4EC9498F@more.net>


These statistics are relative and have no intrisic meaning. What is important is the relative wait time accrued between two periods. You get the raw numbers from one point in time and compare them to another. Statspack, and its crude predecessor utlbstat/utlestat work that way.

They produce nice reports that order wait times and make it easy to spot the worst bottleneck. Which, by the way, is how to tune the database: find the worst bottleneck and relieve it.

Go to Metalink and check out the performance related papers. There is a new white paper that introduces a comprehensive performance management methodology.

Harry Boswell wrote:
>
> I've done a Google search for guidelines on this, but I haven't found clear
> answers. When I check my production instance using TOAD's Server Stats, I
> see some wait stats that seem high, but are they really high? For example:
>
> db file scattered read 16264
> db_sequential_file_read 4642169
>
> dispatcher timer 4915441
>
> latch free 101389
>
> pmon timer 152556
>
> SQL*Net message from client 2429529
> SQL*Net message to client 2429496
> SQL*Net more data from client 65029
> SQL*Net more data to client 1491867
>
> virtual circuit status 2278288
>
> Are these numbers high, or is a high number to be expected?
>
> Thanks,
> Harry
Received on Thu Jun 13 2002 - 21:21:19 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US