Re: Deactivating Redo Logs on 9i?

Ganesh Raja wrote:
>
> Connor McDonald <connor_mcdonald_at_yahoo.com> wrote in message news:<3C7D5D43.6A4_at_yahoo.com>...
> > Edzard wrote:
> > >
> > > Hello,
> > >
> > > Have a look at this extract from the alert log:
> > >
> > > Wed Feb 27 15:06:14 2002
> > > Starting up ORACLE RDBMS Version: 9.0.1.2.0.
> > > System parameters with non-default values:
> > > ..
> > > _disable_logging = TRUE
> > > ..
> > > Current log# 1 seq# 28 mem# 0: /u01/oracle/p1/oradata/redo01.ln
> > > ..
> > > Wed Feb 27 15:27:59 2002
> > > Thread 1 advanced to log sequence 29
> > > Current log# 2 seq# 29 mem# 0: /u01/oracle/p1/oradata/redo02.ln
> > >
> > > What I was doing at the time is measuring the effect on the wait time
> > > of a data processing job:
> > > Whithout _disable_logging: Elapsed: 00:04:22.52
> > > Whith _disable_logging: Elapsed: 00:04:24.05
> > >
> > > Has anyone else measured this, I mean that it makes no difference? May
> > > be it is only like this since Oracle9i.
> > >
> > > Edzard
> >
> > Redo logging still "happens" - just the actual file writing does not
> > take place. So you get log switches, checkpoints etc etc.
>
> Conor Can u pls Beef up on this a Bit more... What do u mean by the above stmt.
>
> If you want
> > to use _disable_logging (ugh!) then you really need to create massive
> > log files to get the "benefits"
> >
> > hth
> > connor
>
> TIA.
>
> Regards,
> Ganesh R

My understanding of how this works is that the log buffer etc is still filled in the normal way, and commits etc invoke LGWR into action. It clears the log buffer, does the all the normal latching you would expect, but just the actual write-to-redo-log call does not take place. So the redo logs still "fill" and "switch" at the antipicated time (and you get checkpoints at the switch).

hth
connor

-- 
==============================
Connor McDonald

http://www.oracledba.co.uk

"Some days you're the pigeon, some days you're the statue..."