Re: dbms_shared_pool.keep

From: Shakespeare <whatsin_at_xs4all.nl>
Date: Sat, 16 May 2009 00:22:02 +0200
Message-ID: <4a0deb12$0$194$e4fe514c_at_news.xs4all.nl>



Matthias Hoys schreef:
> "Shakespeare" <whatsin_at_xs4all.nl> wrote in message 
> news:4a0dbedc$0$195$e4fe514c_at_news.xs4all.nl...

>> Mladen Gogala schreef:
>>> Today, I noticed a performance problem on one of my databases. Statspack
>>> was showing quite a few of library cache waits. Most of the waiting
>>> sessions were trying to execute the same procedure and the problem was
>>> alleviated to significant degree by pinning the the package that contains
>>> the offending procedure, as well as all of the other PL/SQL objects
>>> belonging to FLOWS_03000. Trouble is, I distinctly remember rading an
>>> article claiming that using DBMS_SHARED_POOL.KEEP is not good, that it
>>> is, in fact, obsolete. I can't
>>> locate the article so I will ask anyone who thinks that I shouldn't have
>>> done what I did, to explain me why? In other words, am I missing
>>> something
>>> or not? I have sufficient free memory, no danger of 4031 in the near
>>> future.
>>>
>>>
>>>
>> AFAIK, pinning those packages is a well known way to speed up APEX.
>>
>> IIWIW
>>
>> Shakespeare
> 
> I also remember "something" about DBMS_SHARED_POOL.KEEP being of no more use 
> for 10g.
> I didn't know this was a known trick to speed up APEX, although I can't 
> remember reading about this on the OTN forum. Something to try out next 
> week.
> 
> Matthias
> 
> 

See for example
http://beyondoracle.wordpress.com/2008/07/16/speed-up-execution-of-apex-packages-pinning/

Shakespeare Received on Fri May 15 2009 - 17:22:02 CDT

Original text of this message