Re: Need help understanding a performance problem

On Fri, 18 Jul 2003 23:49:42 +0200, Sybrand Bakker <gooiditweg_at_sybrandb.demon.nl> wrote:

>On Fri, 18 Jul 2003 21:24:19 GMT, hboswel1_at_bellsouth.net (Harry
>Boswell) wrote:
>
>>On Thu, 17 Jul 2003 21:56:53 +0200, "Anton Buijs"
>><remove_aammbuijs_at_xs4all.nl> wrote:
>>
>>
>>>Check the same indexes are created in test as in the production instance.
>>>Check important init.ora parameters like db_block_buffers (can be lower on
>>>test because there are less users and less memory but not too small),
>>
>>Since db_block_buffers was the easiest thing to try, I increased that
>>parameter, from 500 to 2500. The effect was significant - the ratio
>>of sequential reads to scattered reads droppped from 15-1 to about
>>4-1. And response time improved similarly. I think I intended for
>>that parameter to be 5000 anyway - my production db value is 12500.
>>So this whole thing was most likely brought about by a mis-type.
>>
>>Thanks,
>>Harry Boswell
>>
>
>
>I see you are observing the Niemic tuning faith. Likely this will
>bring you nowhere
>

Ah, but I could be persuaded to become a Bakkerite, were I only given words of wisdom!!

IOW, in all seriousness, what's the problem with the increase in db_block_buffers?

Thanks,
Harry Boswell Received on Mon Jul 21 2003 - 15:36:49 CDT