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

Home -> Community -> Mailing Lists -> Oracle-L -> Re: Prod on Solaris and Dr on Linux

Re: Prod on Solaris and Dr on Linux

From: Mark Brinsmead <mark.brinsmead_at_shaw.ca>
Date: Mon, 16 Jan 2006 15:01:33 -0700
Message-id: <43CC17BD.6000901@shaw.ca>

I'm no longer current with the "theory" of Disaster Recovery, but the last time I worked near it, the main rule seemed to be this:

    *Full Disaster Recovery must be completed within 96 hours of the     initial failure. If you can't complete DR within 96 hours don't     bother -- your bankruptcy is now inevitable*.

That was many years ago (much pre-9/11), so standards may be considerably tighter now. Anyway 96 hours (4 days) sounds like a lot of time, but it is not. It includes the time needed to decide to initiate disaster recovery, time to transport backup tapes, time to setup hardware, time to reconfigure networks, time to restore data from backups, etc. And all of this may need to done by somebody who has never seen the original systems.

The first rule of Disaster Recovery is to keep it simple. Platform changes are not simple. They normally require several weeks (or months) of planning and testing.

In a perfect world, your DR hardware (and OS, and software, and...) will be *identical* to production. Sadly, we rarely get this level of luxury. Last time I want through this process (for a large RAC instance with raw storage carefully placed on 1200 disks) I could not even get an assurance of having the same *number* of disks, let alone the same size or type. Having to find a way to automatically re-distribute my raw partitions across an unknown number of disks of unknown size was *hardly* pleasurable. I wouldn't want to contemplate how much more difficult this may have been if I also needed to worry about changes to operating systems. How many shell scripts and batch jobs will fail when I move stuff from Solaris to Linux? What about OS accounts, kernel settings, cron jobs, custom code and libraries, ...

While you may not be able to get *identical* equipment at your DR site, you need to take care to minimize the differences, and to ensure that any differences you can't eliminate will be as transparent as possible.

No, I do not think it is wise to attempt a platform change during disaster recovery. Or more precisely, doing so would not be the result of what I would consider to be "good planning".

Sanjay Mishra wrote:

> Another Awkward question
>
> Can I create mine DR DB on Linux while mine Prod is on Solaris
>
> TIA
> Sanjay
>
> ------------------------------------------------------------------------
> Yahoo! Photos - Showcase holiday pictures in hardcover
> Photo Books
> <http://us.rd.yahoo.com/mail_us/taglines/photobooks/*http://pa.yahoo.com/*http://us.rd.yahoo.com/mail_us/taglines/photos/evt=38088/*http://pg.photos.yahoo.com/ph//page?.file=photobook_splash.html>.
> You design it and we'll bind it!

--
http://www.freelists.org/webpage/oracle-l
Received on Mon Jan 16 2006 - 16:01:55 CST

Original text of this message

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