Re: Help with statspack report comparison

From: Jerome Vitalis <vitalismanN05P4M_at_gmail.com.invalid>
Date: 23 Sep 2007 17:14:25 GMT
Message-ID: <46f69ef1$0$23098$79c14f64@nan-newsreader-05.noos.net>

On Sun, 23 Sep 2007 13:16:14 +0200, Martin T. wrote:

> *** AT10 ***
> Buffer Cache: 48M Std Block Size: 8K
>
> *** DE30 ***
> Buffer Cache: 24M Std Block Size: 8K

> *** AT10 ***
> ~~~~~~~~~~~~ Per Second Per Transaction
> ---------------
> ---------------
> Logical reads: 22,854.65
> 2,039.62 Block changes: 49.10
> 4.38
> Physical reads: 197.40
> 17.62

> *** DE30 ***
> ~~~~~~~~~~~~ Per Second Per Transaction
> ---------------
> ---------------
> Logical reads: 52,318.99
> 22,518.45 Block changes: 20.08
> 8.64
> Physical reads: 1,749.80
> 753.12

In addition to Jonathan and Charles's analysis that will surely help find the major culprit, there might be a buffer cache size issue as well. Is there an actual reason why on the bad box the cache is twice as small as on the good one?
What does the cache advisory have to say about that? Received on Sun Sep 23 2007 - 12:14:25 CDT