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: RAC in NAS

Re: RAC in NAS

From: Mark Brinsmead <pythianbrinsmead_at_gmail.com>
Date: Wed, 26 Jul 2006 19:10:03 -0600
Message-ID: <cf3341710607261810r5b1870f2p8bef6ecab86f5c94@mail.gmail.com>


NAS is "cutting edge"? Well, maybe if you're talking to a sales rep from NetApp, it is. ;-)

As for EE vs. SE, well, the costomer decided that issue for themselves when they decided to build a cluster with 8 CPUs. That makes EE mandatory.

A more relevant question to ask is whether the customer actually *needs* RAC. Or 8 CPUs. If they had not done enough research to know whether the storage layer they had chosen would even work, then chances are, they have not sized the hardware nor adequately researched the requirements for RAC, either.

What are you expected to get from RAC, anyway? I hope they were looking for "high availability" (but then, why only two nodes?) rather than "horizontal scaling".

To answer the original question (assuming nobody else hase yet -- I haven't finished reading the thread), I am reasonably sure that NAS storage *is* supported with RAC, but only two or three specific hardware solutions (NetApp being one), and only within very strict limitations. In fact, among those limitations are "NFS3 or better, over TCP". No UDP. And no CIFS, and hence no Windows. (Okay, I'm only guessing that this is the only reason that NAS is not supported with Windows-RAC, but I think it's a pretty good guess. According to Metalink, CIFS does not support the file-locking services required by RAC. And somehow, I have trouble picturing Windows clusters with NFS...)

One thing you had better consider if you go down the Linux-NFS-RAC route is Asynch I/O. I just finished butting heads with this on a client's single-intance systems, when asked to diagnose severe performance issues that manifested a few months after migrating from DAS to NFS. (Very) long story made short, the database stopped using Asynch (or Direct) I/O following this change. (The consequences did not manifest immediately, but became *very* apparent with the arrival of a new I/O intensive application.) As it turns out, Asynch I/O -- which is a huge advantage to Oracle, and practically *assumed* in most deployments -- is neither supported nor possible with the combination of Redhat Linux and NFS. At least, not as far as I have been able to determine...

This can have huge implications.

If you're *certain* that you *must* use NAS (NFS), then be certain you do your homework when choosing your other components. (Oops. That's how we got here, isn't it?) Redhat cannot (yet) support Asynch I/O with NFS -- the same may be true of other supported Linux distros, but I'm not certain. Perhaps you could use Solaris? [Ooh! Wouldn't *that* annoy IBM... ;-) ]

--

Cheers,
-- Mark Brinsmead

   Staff DBA,
   The Pythian Group
   http://www.pythian.com/blogs

--
http://www.freelists.org/webpage/oracle-l
Received on Wed Jul 26 2006 - 20:10:03 CDT

Original text of this message

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