Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
![]() |
![]() |
Home -> Community -> Mailing Lists -> Oracle-L -> RE: Multiple Buffer Pool
There is something that it doesn't sound good. When you pin packages, =
you do
it on the shared pool. When you keep tables, you do it on the buffer =
cache.
More specific: code, as PL/SQL, procedures, and SQL, runs using the =
shared
pool as a cache. The data blocks goes to the buffer. Through 8 you can
specify two areas in the buffer cache, known as buffer keep and buffer
recycle.=20
Talking deeply about those things. From my experience and my tests I =
can
advise you testing with small tables and indexes used for application. =
Add
their blocks. And with that in mind set the value of buffer_keep. At =
least
include an overhead of 20%. Also estimate the amount of lru_latches for =
that
buffer. BUT REMEMBER THAT IS TEST AND TRY with these new features. With =
my
advice I can tell you that I always gain some percent of performance.
However your case is different, so test it.
About the shared_pool I always pin packages when I notice shared pool
fragmentation problems. Before this I usually set =
shared_pool_reserved_size
to a value near 10%/15% of the shared_pool_size. And when you realise =
about
that kind of problems: looking at x$kghlu and v$sgastat and at last =
when you
receive a 4031 error.
Regards.
NOTE: Remember that before setting the value of buffer keep or recycle, =
you
must issue the alter table....buffer <keep/recycle> stmt on the table.
> -----Mensaje original-----
> De: Gautam_Reddy_at_Dell.com [SMTP:Gautam_Reddy_at_Dell.com]
> Enviado el: Mi=E9rcoles 19 de Julio de 2000 12:39
> Para: Multiple recipients of list ORACLE-L
> Asunto: Multiple Buffer Pool
>=20
![]() |
![]() |