Interesting point about checkpoints, I have not looked into that. For your other questions, under heavy load they can fill up a gigabyte of redo in 2 to 3 minutes, but 5 to 10 minutes is more steady state. The database supports a web-facing application, so outside customer demand can drive traffic much higher than average. Log buffer is set to 14mb. ________________________________ From: David Barbour [mailto:david.barbour1@xxxxxxxxx] Sent: Tuesday, April 07, 2009 2:22 PM To: Crisler, Jon Cc: Rajeev Prabhakar; Oracle-L Freelists Subject: Re: Mirroring redo log groups or not ? Mirroring protects you from more than disk failure, it also protects you from human error. Please, mirror your logs. There is a long history littered with un-employed DBAs backing this recommendation. For your specific problem, you might want to look a a couple of things. How long does it take you to fill up a gigabyte of redo? How long does it take you to write an archive log that's a gigabyte in size? What is your log_buffer parameter set to? Have you evaluated the effect of checkpoints in your situation? On Tue, Apr 7, 2009 at 1:14 PM, Crisler, Jon <Jon.Crisler@xxxxxxx> wrote: The redo logs are 6 groups, 2 members each, 1gb in size. The SAN is EMC Clariion. Raid 5 does have somewhat poorer performance in writes in our environment but its usually not a major factor. -----Original Message----- From: Rajeev Prabhakar [mailto:rprabha01@xxxxxxxxx] Sent: Tuesday, April 07, 2009 12:48 PM To: Crisler, Jon Cc: Oracle-L Freelists Subject: Re: Mirroring redo log groups or not ? Hello Jon, Before you go into redesign of the redo logs, did you check w/the SAs/Storage Admins ? During my last interactions with type of performance issue, I found that RAID 5 had a negative effect on overall performance during QA testing. of course, your mileage may vary. Can you share your findings (as seen by SA/Storage Admins). How big are your redo logs ? What kind of disks (Storage/SAN) are you folks using ? -Rajeev On Tue, Apr 7, 2009 at 12:16 PM, Crisler, Jon <Jon.Crisler@xxxxxxx> wrote: > So, if the LUN that holds the online redo logs is already protected (raid 1, > 5 etc) do I really need to mirror the online redo logs ? Should I convert > the LUN to Raid 1 / 0 ? if I convert to Raid 1/ 0 should I keep redo log > mirroring and point to the same LUN ? -- //www.freelists.org/webpage/oracle-l