Re: Redo log groups

tmgn wrote:

> Hi there,
> Additional Redolog Members( Mirrors) are required just for additional
> Protection, so that LGWR continues to write even if any member (Redo Log file)
> is unavailabe.
> You can also Mirror the Redo Log files at the Hardware level only.
> Having too many Members might slow down the LGWR process as it needs to write
> to all of them in parallel. I would suggest just 2 or 3 members for a REdo Log
> Group.
>
> The Number and Size of a Redolog Group should depend on the Transaction
> volume.This becomes critical if the Database is operated in ARCHIVELOG mode
> becos it is important for the LGWR process to continue writing to the RedoLog
> files without having to wait for the ARCH process to complete copying the
> Filled Online redolog file to the LOG_ARCH_DEST location.
>
> If you encounter any messages like 'cannot allocate log' in the alert.log file
> then you can increase the number of Redolog files providing more room for LGWR
> to perform transaction logging and time for ARCH process to complete the Copy.
>
> You can also increase the Size of the Log file if the Log Switches are
> happening too rapidly. It should be tuned so that approx 2/3 switches happen in
> any 1 hour time interval.
>
> Hope this note helps..
>
> -Thiru
> Joao Ricardo B Oliveira wrote:
>
> > hi there:
> >
> > Anyone can explain when we need to create more Redo Log members or Redo
> > Log groups ?
> >
> > Another question is :
> >
> > If i submit several jobs in the OEM, like this :
> >
> > 1.broadcast message 'Shutdown in 10 minutes'
> > 2.shutdown immediate everyday at 18:00
> > 3.backup database when database down at 18:10
> >
> > if one of them fails, for example the shutdown, the backup also
> > fails ?
> >
> > if i want to backup directly to a tape (HP Surestore Dat8) the backup
> > manager asks for the Tape Device ID... where can i find it ?
> >
> > Thankx
> >
> > jricardo_at_ipb.pt
> >
> > "Oracle the edge @ side ..." by me...

From my understanding, you should not mirror the redo log files at the hardware level only. When the log writer process determines that it doesn't know the validity of the contents of a particular redo log, it will mark it as STALE in the v$log table. If this redo log is the only copy, then it can't be archived, and will cause a database halt (in archivelog mode). If the logs are mirrored at the Oracle level, then the archiver process will choose a copy of the redo log not marked as stale, so no interruptions. If mirrored only at the hardware level, then both copies will be marked as STALE. Received on Tue Jul 20 1999 - 19:58:01 CDT