Re: Will this configuration for RAC work properly?

From: Marc Bruinsma <marc.bruinsma_at_chello.nl>
Date: Mon, 01 Oct 2007 11:14:31 +0200
Message-ID: <ZT2Mi.3518$Hv4.995@amstwist00>

charlie cs wrote:

> 
> "DA Morgan" <damorgan_at_psoug.org> wrote in message
> news:1191019381.731621_at_bubbleator.drizzle.com...

>> charlie cs wrote:
>>> Dear Group,
>>>
>>> We are trying to setup a RAC environment in our shop. One proposal is
>>> like this:
>>>
>>> We have node A1, A2, B1, B2, and san box S1,S2
>>>
>>> A1,A2, S1 is on primary site, B1,B2,S2 is on disaster recovery site,
>>>
>>> A1,A2 is connecting to S1, B1,B2 is connecting to S2, we will connect
>>> A1,A2,B1,B2 through fiber glass chennel and S1,S2 are synchronized. In
>>> this way, we can put A1,A2,B1,B2 all in service, because S1,S2 holds
>>> same datebase(two databases but synchornized)
>>>
>>> Any comments?
>>>
>>> Thanks for your help.
>>
>> How many miles between data center AS1 and data center BS2? Look
>> up the details for extended RAC clusters and see if you qualify?
>>
>> What version of Oracle? It matters.
>>
>> Systems like this have been built but not the way you are describing
>> it. A third site, generally NFS mount or iSCSI, is required to hold the
>> voting disk and thus you need two separate fibre networks.
>> --
>> Daniel A. Morgan
>> University of Washington
>> damorgan_at_x.washington.edu (replace x with u to respond)
>> Puget Sound Oracle Users Group
>> www.psoug.org

> 
> Oracle is 10.2.  The location between two site is about 2 miles away.
> Basically we do not like one or several nodes sitting there doing nothing,
> we like both DR site and primary site share the work load.
> 
> Any more help?
> 
> Thanks very much

There is one question that in my opinion is still unanswered here, and that is: are you talking about a 4-node extended cluster with either host- or array based mirroring between the SAN's or are you talking about 2-node RAC-to-2-node RAC Dataguard?

Big difference.

If it's an extended cluster you are talking about, then you will need a third voting disk on a separate node somewhere, could be a NetApp filer, or even a Linux box with an NFS mount. I've built a few of those with ASM mirroring, and it works fine. That will give you four nodes to play with.

However, I suspect you are talking about RAC-to-RAC dataguard, since you refer to a disaster recovery site, and depending on whether it is a physical or a logical standby, you could use the standby for things like reporting or backup purposes, but with certain restrictions.

Please read up on the Dataguard docs and what you can and cannot do with your standby, because there are several ways to set it up, and you can find yourself in deep s**t if it doesn't do what you hoped it would do.

Marc Received on Mon Oct 01 2007 - 04:14:31 CDT