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

Home -> Community -> Usenet -> c.d.o.server -> Re: Raid 5 vs mirroring

Re: Raid 5 vs mirroring

From: Steve Phelan <stevep_at_XXnospamXX.toneline.demon.co.uk>
Date: 1998/03/12
Message-ID: <889718110.29124.0.nnrp-08.c2de712e@news.demon.co.uk>#1/1

soverby_at_nspr.com wrote in message <6e8rt6$ed6$1_at_nnrp1.dejanews.com>...
>I don't have a direct comparison of SSA mirroring to SSA RAID5, but
> I did setup a tablespace on 2 SCSI drives with the following results:
>
>Import 2.7 million row
>no indexes in place
>space pre-allocated
>
>SCSI mirrored = 10 min,
>SSA RAID5 c/FWC = 5.3 min

Er, these figures are completely meaningless. An SSA drive array versus a normal SCSI drive pair? It's not a like with like comparison in any way. I'm not knocking your effort or results, just the test in the first place...

The interesting question is what a mirrored and striped SSA drive pair would have achieved in comparison, and in that case I would *expect* the RAID5 solution to be slower, esp. on 'write' performance. And you have to look at both read and write, and the amount of those your database suffers to come up with something meaningful.

Of course, the amount of RAM you had free to buffer in and how I/O bound your system is will also be quite a factor.

It's just such a comparison we are about to undertake with our server.

Steve Phelan. Received on Thu Mar 12 1998 - 00:00:00 CST

Original text of this message

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