Re: Sync redo log waits

From: Andrew Kerber <andrew.kerber_at_gmail.com>
Date: Tue, 19 Feb 2013 12:50:10 -0600
Message-ID: <CAJvnOJYm9-H8xWq9qaDy+HvWWXnRQ5-v7gstN-7WMqm4ZFsXEw_at_mail.gmail.com>

Oracle has a setting, archive_lag_target, that will force a log switch if one has not been done in the time frame specified by archive_lag_target. So when the redo activity is uneven, I usually set that parameter and set the redo logs to a very large size.
On Tue, Feb 19, 2013 at 11:48 AM, Paul Houghton < Paul.Houghton_at_admin.cam.ac.uk> wrote:

> Hi
>
> According to a number of oracle support notes, e.g. 147468.1 performance
> will be impacted if log switches more often than about once every 15
> minutes. That note specifically says that performance will be impacted by
> switching once every 3 minutes, which is around what you are doing.
>
> So yes, you should increase the size of the redo logs to 4.5 - 6 times
> their current size unless you are sure you can reduce the number of
> switches by reducing the workload.
>
> We have a cron job which switches the log file every 15 minutes to ensure
> the standby is up to date, though I am not sure if this is needed any more
> (11.2).
>
> You may be able to talk with the application people to get them to spread
> the load. Sometimes you get a trendy time to start a batch process (say
> 2am) and everyone starts their process at this time, though I can't see a
> pattern in the data you posted. Does the application keep some kind of log
> or audit of what it was doing which you could consult to find out what was
> being run, or use AWR reports if you have tuning pack, or even statspack
> and tune that area of the application or move it to a quieter time?
>
> Hope this helps
>
> PaulH
>
>
> --
> http://www.freelists.org/webpage/oracle-l
>
>
>

-- 
Andrew W. Kerber

'If at first you dont succeed, dont take up skydiving.'


--
http://www.freelists.org/webpage/oracle-l

Received on Tue Feb 19 2013 - 19:50:10 CET