Re: DBWR Tuning

From: DA Morgan <damorgan_at_x.washington.edu>
Date: Tue, 14 Dec 2004 16:19:05 -0800
Message-ID: <1103069832.171865@yasure>

Niall Litchfield wrote:

> "DA Morgan" <damorgan_at_x.washington.edu> wrote in message
> news:1102988750.978267_at_yasure...
>

>>I'm fine with it dying but want to say that, from experience,
>>there is no hard and much good that comes from experimentation.
>>All guidelines are just that ... guidelines. 8 is not more a
>>"fact" than is any other integer. It is always best to test
>>modifications to configurations.

>
>
> I'll agree with all of that, and I'm a bit wary that it sounds like I'm
> having the last word. I'd just like to be on record as saying that *in
> general* you don't need one dbwr per cpu, not that you must wait until you
> have 8 cpus. Once again I'd be driven by the percieved performance of the
> system for the users. It seems to me that the performance tuning our job as
> dbas to translate 'my order entry screen is too slow' into specific
> technical steps, not to continually perform technical adjustments, because
> they 'seem' appropriate.

I'll rescue you from having the last word. I just read the mangled prose I wrote and ... I've got to start proof-reading after when I type with single-malt in hand.

-- 
Daniel A. Morgan
University of Washington
damorgan_at_x.washington.edu
(replace 'x' with 'u' to respond)

Received on Tue Dec 14 2004 - 18:19:05 CST