Re: tuning redo logs

In article <01be4042$37a60940$9f0a13ac_at_DAS.mincom.co.id>, "Daniel Ady Surya" <Daniel.Surya_at_mincom.co.id> wrote:
>Hi Robles..,
>The online redo log files record the transactions that occur within
>database.
>We have to prevent a log switch occured no more often than every 30
>minutes.
>Use this query to show the time of each log switch:
>Select to_char(to_date(time,'MM/DD/YY HH24:MI:SS'),'HH24:MI:SS')
>from v$log_history where to_date(time,'MM/DD/YY HH24:MI:SS') >
>sysdate-1 order by to_date(time,'MM/DD/YY HH24:MI:SS') desc;
>If log switch occured more often than every 30 minutes, the online redo log
>files are too small.

I disagree on this point. The rule of thumb is a log switch every 15 minutes, but it isn't carved in stone. The tradeoffs are archiver arctivity v.s. instance recovery time and of small frequent v.s. larger, less frequent archiver writes (essentially a file copy).

>Be aware, if large online redo log files are used, instance recovery will
>take a long
>time because of the large number of redo log entries that must be applied.
>
>I hope this will help you
>Regards,
>
>Daniel AS
>IT Consultant (DBA)
>
>Osvaldo Robles <orobles1_at_coqui.ney> wrote in article
><2Vyn2.305$Za3.2692_at_news13.ispnews.com>...
>> Hi everyone
>> I would like to know if there is a formula or a method to determine
>the
>> number of redo logs necessary for a database that handles about 16
>million
>> records.
>>
>> thanks in advance

• OraSaurus --
• Remove "not_" to reply... --