Re: Why extra standby redo log group?
Date: Sun, 5 Apr 2009 09:19:23 -0700 (PDT)
Here is an example form one of my standby databases GROUP#,DBID,THREAD#,SEQUENCE#,BYTES,USED,ARCHIVED,STATUS,FIRST_CHANGE#,FIRST_TIME,LAST_CHANGE#,LAST_TIME
5,2993939164,1,539739,1258291200,512,YES,ACTIVE,124541122419,4/5/2009 11:15:20 AM,124541122522,4/5/2009 11:15:20 AM 6,2993939164,1,539738,1258291200,294810112,NO,ACTIVE,124540991837,4/5/2009 11:03:59 AM,124541122419,4/5/2009 11:15:20 AM 7,UNASSIGNED,1,0,1258291200,512,NO,UNASSIGNED,0,,0, 8,UNASSIGNED,1,0,1258291200,512,NO,UNASSIGNED,0,,0, 9,UNASSIGNED,1,0,1258291200,512,NO,UNASSIGNED,0,,0,
As you can see both are active but one is archived=yes the other is not sicne it is a realtime recieve . in a very busy environment you will see all unassigned as Active with one being used as realtime till time that the backlog is more than the standby redo logs.
- Original Message ---- From: Yong Huang <yong321_at_yahoo.com> To: john.hallas_at_morrisonsplc.co.uk; fuadar_at_yahoo.com; Martin Brown <martinfbrown_at_hotmail.com> Cc: oracle-l_at_freelists.org Sent: Sunday, April 5, 2009 9:39:09 AM Subject: RE: Why extra standby redo log group?
Martin, John, Fuad,
Thank you. When you have heavy transaction for a period of time, if you have n log groups on primary and n+1 SRL groups on standby, do you see all n+1 SRL groups used? That is, on the standby, do you see all, not just n, rows in v$standby_log under status column, alternately showing 'ACTIVE'?
I know if I manually switch logfile on primary, only n SRL groups will be 'ACTIVE' (the extra stays unused). I'll test by building a small data guard where primary is on a node with fast storage and standby with slow storage and create lots of redo.
- On Sun, 4/5/09, Martin Brown <martinfbrown_at_hotmail.com> wrote:
From: Martin Brown <martinfbrown_at_hotmail.com>
Subject: RE: Why extra standby redo log group?
To: john.hallas_at_morrisonsplc.co.uk, fuadar_at_yahoo.com, yong321_at_yahoo.com
Date: Sunday, April 5, 2009, 8:46 AM
Totally agree. Our configuration has 8 primary nodes and (of course) only 1 standby node. For those that don't run DataGuard 10g, you can only have 1 active standby node. Normal log switches happen about 3 per hour. During peak times, our log switchs pick up speed and this configuration keeps up quite nicely.
> From: John.Hallas_at_morrisonsplc.co.uk
> To: fuadar_at_yahoo.com; yong321_at_yahoo.com
> CC: oracle-l_at_freelists.org
> Date: Sun, 5 Apr 2009 14:33:16 +0100
> Subject: RE: Why extra standby redo log group?
> I would agree with what Fuad says, it is to ensure
that the standby can keep up with the primary. It is only a recommendation though and not mandatory
> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Fuad Arshad
> Sent: 04 April 2009 03:44
> To: yong321_at_yahoo.com
> Cc: oracle-l_at_freelists.org
> Subject: Re: Why extra standby redo log group?
> Well in my case I've seem standby redo logs used
to cover for backlogs I.e Log switching is faster than the standby instance can perform