Noons wrote:
> DA Morgan wrote:
>
>>But why not just move everything to RAW and ASM?
>>
>>I've seen no problems with in on RH 3.
>>--
>
>
> Because there is a limit on the number of raw
> devices one can open in RH3 and I'm w-a-a-a-y
> over that already in my db servers.
>
> And because I can't convince my management to
> buy upfront all the disk storage we may need
> for the next 5 years. It changes so much in price
> as time goes by, anyone buying more than 12 months
> ahead is basically throwing away good moolah.
>
> Therefore I have to live with adding space to existing
> dbs as time and needs come by.
> ALTER DATABASE DATAFILE '/dev/blah' RESIZE 40000M
> ain't work well when /dev/blah is raw. The alternative
> of changing via lvm ain't work either if you're 24X7!
>
> And don't get me started on how many raw devices
> I need to run my redos!
>
> Apple's XSERVE raids have been a Godsend: good
> performance even in cooked IO, cheap, large.
> But even these have limits.
>
> All I need is ONE 9irn that works well...
Take a bunch of disks.
Stripe them.
Present them as a single LUN.
One raw device.
--
Daniel A. Morgan
http://www.psoug.org
damorgan_at_x.washington.edu
(replace x with u to respond)
Received on Fri Jun 24 2005 - 00:05:25 CDT