Re: Cache Hit Ratio from system views

From: DA Morgan <damorgan_at_psoug.org>
Date: Wed, 08 Aug 2007 13:14:29 -0700
Message-ID: <1186604067.44579@bubbleator.drizzle.com>

Niall Litchfield wrote:
> sybrandb_at_hccnet.nl wrote:

>> On Wed, 08 Aug 2007 12:02:46 -0700, Teresa Masino
>> <teresa.masino_at_peninsula.org> wrote:
>>
>>> Alone, yea, but they can sometimes be an indicator that an instance
>>> needs attention.
>> Do you really think so? Yesterday I had a customer spouting flames
>> because of a RAC database coming down to a crawl due to buffer busy
>> waits in  conjunction with both instances updating the same table over
>> and over again.
>> Nothing to be seen from the BCHR.
>>
>> But then of course BCHR is a MIPS (Meaningless Indicator of the
>> Performance of the System). 
>> Only good old Don Burleson makes people erroneously believe they can
>> tune their database by throwing memory at the problem to crank up the
>> BCHR.
>> Savy professionals (a favorite phrase of Don) know better.
>>

>
> enough of you will recognise where I stand on ratios. Indeed I believe
> that I presented on the effectiveness or in fact otherwise of the BCHR
> on its own nearly 5 years ago.
>
> I have some sympathy withe Teresa's sentiment above though. Imagine a
> system that has had a truly horrible burleson concerned hopelessness
> ratio of 75% since 2002. Today the BCHR is 25% (or indeed 98%). I
> venture to suggest that it's a change worth determining the reasons for,
> either way. Ratios don't mean anything much, abrupt changes in them
> often do.

On the other hand monitoring things that matter would save you the trouble of chasing after smoke in a house of mirrors.

There are real things that can be monitored. And if Theresa has the license to use AWR as it seems she'd be far better served looking at things that matter.

-- 
Daniel A. Morgan
University of Washington
damorgan_at_x.washington.edu (replace x with u to respond)
Puget Sound Oracle Users Group
www.psoug.org

Received on Wed Aug 08 2007 - 15:14:29 CDT