Re: Oracle slowed down

From: Bob Jones <email_at_me.not>
Date: Sat, 22 Apr 2006 14:38:04 GMT
Message-ID: <gdr2g.62832$F_3.48771@newssvr29.news.prodigy.net>

"Frank van Bortel" <frank.van.bortel_at_gmail.com> wrote in message news:e2dbid$i9k$1_at_news3.zwoll1.ov.home.nl...
> Bob Jones wrote:

>> "HansF" <News.Hans_at_telus.net> wrote in message
>> news:pan.2006.04.22.01.32.46.787609_at_telus.net...
>>> On Sat, 22 Apr 2006 01:05:09 +0000, Bob Jones wrote:
>>>
>>>>> A 100% BCHR can just as easily identify a grossly overconfigured cache
>>>>> as it can identify popular data sets.  In both cases the
>>>>> misinterpretation is that it looks good.
>>>>>
>>>>>
>>>> That's why we need to tune the buffer cache.
>>> Tuning BCHR is NOT tuning buffer cache.
>>>
>>> Buffer Cache is important, not BCHR.
>>>
>>> <sigh>
>>>
>>
>> No, you do not tune BCHR.  It's only there to help you tune buffer cache.
>> How do you know buffer cache needs tuning without looking at BHCR?
>>
>>
>

> Because you look at the waits.
> Any waits. Maybe you need to tune the buffer cache, maybe
> you need to kick the sysadmin's ass.

What waits?

> Whatever.
> You want a BCHR of 30% - I'll get it for you.
> You want 100% -I'll get it for you.
> And in the mean time, in both cases, end users are crying,
> and tearing their hair out.
> --

Sure, have you ever seen a system with 30% performing better than 100%, provided everything else being equal other than the buffer cache. Received on Sat Apr 22 2006 - 09:38:04 CDT