Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: 10g/AIX: Database Buffers problem
"Victor Oosterbaan" <voosterbaan_at_desyde.nl> wrote in message
news:92f6cf14.0406100253.6e2197ec_at_posting.google.com...
> "Howard J. Rogers" <hjr_at_dizwell.com> wrote in message
news:<40c62238$0$8988$afc38c87_at_news.optusnet.com.au>...
> > "Alvaro Fuentes" <alvarof2_at_hotmail.com> wrote in message
> > news:ca55ob$i14$1_at_ausnews.austin.ibm.com...
> > >
> > > Fellow Oracle Users:
> > >
> > >
> > > I am running Oracle 10g Server on AIX 5.2
> > >
> > > When I try to start my instance, it comes with:
> > >
> > >
> > > SQL*Plus: Release 10.1.0.2.0 - Production on Tue Jun 8 14:43:29 2004
> > >
> > > Copyright (c) 1982, 2004, Oracle. All rights reserved.
> > >
> > > SQL> Connected to an idle instance.
> > > SQL> ORACLE instance started.
> > >
> > > Total System Global Area 2550136832 bytes
> > > Fixed Size 1328064 bytes
> > > Variable Size 2523642944 bytes
> > > Database Buffers 0 bytes
> > > Redo Buffers 25165824 bytes
> > > Database mounted.
> > > Database opened.
> > > SQL> Disconnected from Oracle Database 10g Release 10.1.0.2.0 - 64bit
> > > Production
> > >
> > >
> > > What parameter in my pfile should I set to allocate
> > > Database_buffers?
> >
> >
> >......
> >
> > Bear in mind, that's an awfully big SGA_MAX_SIZE you appear to have
(2.5GB).
> > And no-one, but no-one, needs 25MB for LOG_BUFFER. No Log Buffer needs
to be
> > bigger than, oh about 6MB at the absolute outside. Most people can get
away
> > with 1 or 2MB tops.
> >
> > Little question here; > Why don't we need more then 6 MB of LOG_BUFFER? > Any docs or thoughts on that?
Because the nature of redo is that it is small and that it is flushed very regularly (every 1MB or every 1/3rd of the log buffer). You aren't trying to hold onto redo in memory as your are data buffers or execution plans. The aim is to get it down to disk pronto. So what need of an enormous log buffer?
6MB is not hard and fast, but I've yet to meet a database that needed bigger than that.
As ever, it's measuring waits (redo allocation retries, or the log buffer space event) that tells you whether you do or not.
Regards
HJR
Received on Thu Jun 10 2004 - 07:40:34 CDT