Re: How to resize redo logs

They should be as big as are needed to get log switches happening at the "right" interval.

Huge log files will take ages to fill, and you will therefore log switch infrequently. Since there is a major checkpoint at each log switch, that means you avoid having one hell of an i/o fest. Performance therefore goes up a bit. Since Instance Recovery means Oracle replays all transactions since the last checkpoint, however, then huge log files mean longer Instance Recoveries.

Tiny log files will fill rapidly, and will therefore switch like crazy -lots of checkpoints, poor performance, but good Instance Recovery times.

Your mission is to find somewhere along that spectrum where you can be happy. The usual 'rule of thumb' that a lot of DBAs adopt is to arrange for the files to be big enough that they switch every half-hour to an hour, on average. But that may well not be what is suitable in your particular case.

You can check how often you are log switching by checking the Alert Log. Messages such as "Thread 1 advanced to Log Sequence 15634" indicate a log switch.

If you ever see a message to the effect that "Thread 1 unable to advance to Log Sequence 15635", then it means you haven't got *enough* redo log groups (the size of them having little or no bearing on that particular message).

Regards
HJR <MAbbouchi_at_IITRI.ORG> wrote in message
news:DD42667CBCE4D3119D0700104B95123F3016CA_at_abmail.iitri.org...
> How to resize redo logs and is there a special consideration to be aware
 of?
>
> Thanks
> Mo
>
> --
> Posted from nat05.IITRI.ORG [205.167.170.5]
> via Mailgate.ORG Server - http://www.Mailgate.ORG
Received on Thu Feb 01 2001 - 15:02:13 CST