Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.tools -> Re: redo logs and IO please advise

Re: redo logs and IO please advise

From: Dave Francis <d_a_francis_at_yahoo.co.uk>
Date: Sat, 31 Mar 2001 20:44:10 +0100
Message-ID: <Yqqx6.1535$BR5.16833@news2-hme0>

So long as your redo log files change about every half hour, according to standard guidelines, they`re correctly saved. Bear in mind that batch jobs - if you`re running `em during the day - will probably produce more redo log activity.

Redo log statistics don`t really matter - beyond reasonable limits. So long as they`re not writing excessively - in which case, size `em bigger........... remember the point that they`re only used in instance recovery. So they`ve got to be there, that`s all !!! And - not tying your system IO up excessively.

Dave

Piet de Visser wrote in message <99vpqf$1377$1_at_buty.wanadoo.nl>...
>Here is my script (ora8 version) m
>Obvious remarks :
> - depending on disk-system, usage etc... have seen upto 10000.0 kb/s on
>fiberchannel disks.
> - redo is mostly the hottest disk in a oltp or batch system, and almost
>always a hit with tuning.
>
>HTH
>
>doc
> scan04.sql
> Transaction volume for log switches.
> - less then 1000 seconds per switch is "fast",
> consider larger logfiles
> - Last column > 100.00 is considered "high volume"
> examine batches, try to spread (batch) load over time
>#
>
>column sec format 9999
>column ch_per_sec format 99999.999
>column kb_per_sec format 99999.999
>
>select
> to_char ( lh2.first_time, 'DD-MON-YY HH24:MI:SS' ) switch
>, TO_NUMBER ( lh2.first_time
> - lh1.first_time
> ) * 3600 * 24 seconds
>, ( l.bytes / 1024 )
> / ( TO_NUMBER ( lh2.first_time
> - lh1.first_time
> ) * 3600 * 24
> ) kb_per_sec
>from v$log_history lh1
>, v$log_history lh2
>, v$log l
>where lh1.sequence# = lh2.sequence# -1
>and TO_NUMBER ( lh2.first_time
> - lh1.first_time
> ) * 3600 * 24 > 0 /* min. 60 avoids hot bck peaks */
>and l.status = 'CURRENT' /* assume all groups equal file-sizes */
>and lh1.sequence# > l.sequence# - 500
>order by lh1.sequence#
>/
>
>
>--- end of script
>"Kev.-" <java2e_at_yahoo.com> wrote in message
>news:3ac2a0cd.329368987_at_ns1.nothingbutnet.net...
>> Hello All,
>>
>> Now that I know the right question to ask here it goes. What table
>> can you query to get the IO (physical reads and writes) that they are
>> producing?
>>
>> For example when I query v$filestat I am able to get reads and writes
>> that the file has produced since the oracle server was booted as the
>> values keep incrementing until re-booted/ shutdown-re-started, etc...
>>
>> How can I get these same values for the redo logs. I have looked the
>> the v$log, v$logfile but there is nothing there.
>>
>> Thanks,
>>
>> Kev.-
>
>
Received on Sat Mar 31 2001 - 13:44:10 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US