Re: Re: Re: Performance with one Sun T3 disk array

> It is my belief that T3 supports RAID 0,1 or 5.

Do you know that the T3 does not support RAID 1+0 or RAID 0+1 at the hardware (controller) level?

>Apart from that answer, RAID 0+1 is different than RAID 1+0.
 

That had not occurred to me before, and I'm don't see the difference if both levels of RAID are implemented on the same "level" i.e. hardware or OS. For the example of all hardware RAID, if you mirror drives 1 and 2, 3 and 4, and 5 and 6, and then you stripe across the three pairs, the result is essentially the same as if you stripe across 1,2 and 3 and then mirror that volume onto 4,5 and 6, is it not?

RAID 1+0 is different from RAID 0+1 if mirroring is implemented at the hardware level and striping at the OS (or Veritas) level or visa versa. I'm sure there are other examples.

One of the responders to this thread sent me an absolutely terrific white paper from Oracle on this topic. It is titled "Optimal Storage Configuration Made Easy" by Juan Loaiza. It is a very scientific and well reasoned discussion. Here is the link:

http://technet.oracle.com/deploy/availability/pdf/oow2000_sane.pdf

This paper advocates the S.A.M.E. methodology, which stands for Stripe And Mirror Everything. However, in the details they actually propose hardware mirroring at the controller level, and striping on top of that. Shouldn't we really call this M.A.S.E. ? ;-)

John  

-- 
Spam protected message from: 
Sent by john.beckerle from  ps  piece  from ge part from com
Posted via http://www.usenet-replayer.com/cgi/content/new