Re: RAID0+1 Configuration Redo Logs

But if the redo gets busy, that leaves one disc busy for a long time, which means
that any other activity on that disc is
hit.

Since there aren't enough discs to dedicate the redo to their own discs, it is better to stripe.

Your general point though of avoiding a single redo write hitting multiple discs is good; so it would be sensible to work out the likely scale of the commonest redo write size and put that much (rounded to 64K) as the write/per/disk size on the striping you put onto the redo volumes.

NB -- you don't have to stripe everything at the same stripe size. (at least, not with the LVMs I've worked with).

--

Jonathan Lewis
Yet another Oracle-related web site:  http://www.jlcomp.demon.co.uk

Eugene Firyago wrote in message <8abihn$m43$1_at_bob.news.rcn.net>...


>Since redo log files are only written sequentially and never read randomly

>the good idea is to put each of them on separate raw disk slices.

>

>Good luck,

>Eugene.

>

>

>Colin Taberman <ctaberma_at_lhr-sys.dhl.com> wrote in message

>news:8a8911$eqp$1_at_nnrp1.deja.com...

>> I have 43GB Oracle installation which I am looking to stripe on an

>> array. My initial proposal was to use a 8K stripe size to stripe all

>> data and redo logs across 10 x 4.3 GB disks.

>>

>> Is it prudent to stripe redo logs ? what is the general rules to apply

>> here. Any help appreciated.

>>

>> Colin

>> --

>> DHL Systems

>> Europe & Africa

>>

>>

>> Sent via Deja.com http://www.deja.com/

>> Before you buy.

>

>