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: Oracle configuration on Enterprise E450

Re: Oracle configuration on Enterprise E450

From: Darren Dunham <ddunham_at_redwood.taos.com>
Date: Thu, 05 Dec 2002 04:04:28 GMT
Message-ID: <gfAH9.2563$O22.57046074@newssvr13.news.prodigy.com>


In comp.unix.solaris grjohnson <Johnsog123_at_hotmail.com> wrote:
> Do you mean your redo logs groups have only one member and then mirror
> your redo groups at an os level (instead of multiplexing them via
> oracle, i.e. groups have multiple members on separate disks.) If so,
> then you have Single point of failure. Because if a redo log becomes
> corrupt, that corruption is mirrored also.

> When it comes to redo logs, I'd multiplex via Oracle. I know it may
> seem you are wasteing alot of diskspace, but it's owrth it.

Hey, fine by me. Of course I'm not usually paying for the disk, but trying to convince management that what I'm recommending is a good use of their money.

I guess by saying I "prefer" disk mirroring that it was assumed I preferred it to the exclusion of doing application mirroring.

I'm saying I want the disks mirrored at an OS level. Too many places I've worked have run into disk issues. "Oh no! We're out of space. We need to put this data somewhere *now*!". Then they pick the doesn't- -to-be-redundant redo disk. *bam* lose a disk and lose data.

So when the disk dies, if everything is redundant..

  1. I don't have to worry if someone has moved something else into the redo log filesystem.
  2. I don't have to find out if some script has been trying to get statistics on all the filesystems, and now has 36 copies in memory all hung on the failed filesystem.
  3. I don't have to figure out how to forcibly unmount this filesystem on a live machine, recreate, and remount everything.

I'm not about to tell someone that knows far more about database management than I do that having multiple OS copies of the logs is a waste of space just because it's already redundant at a disk level.

I'm saying I don't agree with the other argument that just because there are multiple OS copies of a file, that doesn't mean that the storage the file is on shouldn't be redundant.

-- 
Darren Dunham                                           ddunham_at_taos.com
Unix System Administrator                    Taos - The SysAdmin Company
Got some Dr Pepper?                           San Francisco, CA bay area
         < This line left intentionally blank to confuse you. >
Received on Wed Dec 04 2002 - 22:04:28 CST

Original text of this message

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