Re: Making Oracle on NAS more resilient to network failures

From: Daniel Nichols <daniel_at_NOSPAMrdnichols.com>
Date: Thu, 18 Sep 2003 20:05:29 +0000 (UTC)
Message-ID: <jj1kmvscsil8l5gvr5f3fr0c6ufrcmd445@4ax.com>

On 17 Sep 2003 09:33:34 -0700, norwoodthree_at_my-deja.com (NorwoodThree) wrote:

>The best way to remedy NAS issues, regardless of what vendors or
>Oracle says, is to not NFS mount Oracle datafiles. Consider a SAN or
>locally attached storage.

Not entirely helpful advice there NorwoodThree. (I'm using CIFS, not NFS, by the way.) With the changes I have suggested in my note I think it is possible to protect Oracle from minor network problems.

With regards to what happens within Oracle when we do get network outages I now believe, from a TAR reply I've had, that only the calls that need access to disk wait. Anything else, like queries that can get their data from buffers, are completed. Next time I test this situation I'll monitor V$SESSION_WAIT. Received on Thu Sep 18 2003 - 15:05:29 CDT