Venkat Rao wrote:
> Hi,
>  
>  I had read in some sites and notes that Pinning objects really does not help in performance, can anyone tell me any stats to prove this and wot is the creteria to decide pinning or not and how to find the eligible candadites to find for pinniing?
>  
>  I felt frequently accessed objects should be pinned and anyway they are always almost in memory as they are being frequenlty accessed, why pinning then? apart from 01st time to laod?
>  
>  any links,advices would be highly apprecaited...
>  many thanks in advance...
>  
>  
>  Reg,
>  Venkat
>  
In many real life cases you are right that frequently accessed objects
will stay in the library cache
anyway thus eliminating the need for pinning objects.

The problem arises from the usage of many very large packages (like with
Oracle Applications) -
probably without the possibility to use several GB's for the shared pool.
In this scenario the large packages tends to come first in line to be
thrown out of the library cache
(according to Murphy's law :-).
As they are invariably used again (talking about Oracle Applications)
you will be hit more by the
reloads of these packages than small objects as they take longer to
reload and are more difficult to
get room for.

The metrics to look out for would be reloads in v$librarycache and
waits/latches referering to the
library cache or the shared pool (in v$system_event / v$latch).

The suggestion to pin (large) objects also comes from the idea of
"prewarming" the library cache
so that the load time of the normally used pachages are paid during
startup - and not by the unsuspecting
user that happens to be the first to use a particular packages after
(re)start.

--
Med venlig hilsen

Martin Berg

email: martin@(protected)
web: www.berg-consult.com
mobil: 30 35 58 68


--
To unsubscribe, email: suse-oracle-unsubscribe@(protected)
For additional commands, email: suse-oracle-help@(protected)
Please see http://www.suse.com/oracle/ before posting