Re: RAID-5

Hi Kenneth:

Check out my web-site below. I have a whole section that discusses RAID and Oracle files, plus equations to calculate the theoretical max workload limits for the disk arrays.

Have fun,

--
David C. Sisk
The Unofficial ORACLE on NT site
http://www.ipass.net/~davesisk/oont.htm

Kenneth C Stahl wrote in message <377766CA.C27277CB_at_lucent.com>...
>Looking for a bit of esoteric experience here.
>
>I have a two-processor Compaq Proliant running Unixware 2.1.2 and Oracle
>7.3.4 with two instances running simulaneously for different
>applications. My filesystems are on RAID5. Oracle documentation
>specifically warns against having redo logs on RAID devices. Right now I
>do not have any choice because I'm stuck with this hardware
>architecture. I'm really not seeing any specific problems.
>
>So, my questions are these:
>1. Is there trouble brewing down the line - or does the fact that
>everything runs fine at present mean that Oracle's warning is related to
>performance?
>2. If the only concern is over performance, how would I calculate the
>related metrics. Due to the nature of the the applications the database
>is more or less loafing along. Under normal circumstances my 10M redo
>logs are not switching very often. On one database I see about 2-5
>switches a day and on the other it is about 10-15 per day.
>3. How much would I really gain by adding a SCSI hard drive for the sole
>purpose of having the redo logs defined there?
>
>If I want to make my case to the people with the money and power I need
>to present evidence that would probably be in the "overwheming"
>category. Right now they are content that the system is working as it
>should and are not inclined to make any changes.
>
>I'd welcome any relevant comments.
>
>Ken
>
Received on Mon Jun 28 1999 - 22:08:39 CDT