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: 10g RAC and SRDF

Re: 10g RAC and SRDF

From: hpuxrac <johnbhurley_at_sbcglobal.net>
Date: 4 Oct 2006 14:04:05 -0700
Message-ID: <1159995845.905970.233640@c28g2000cwb.googlegroups.com>

BCB wrote:
> hpuxrac schreef:
>
> > Talat wrote:
> > > All,
> > >
> > > Apologies for the double post.
> > >
> > > I've just had confirmation from someone within Oracle that not
> > > replicating the binaries and OCR to the DR nodes is the way to go.
> > >
> > > Cheers,
> > >
> > > T.
> >
> > That makes sense to me. We have a similar environment except hpux and
> > 9.2 ( so far ).
> >
> > The machine that in a disaster would host production database runs the
> > test database normally. We also have a separate install of oracle on
> > both machines ...
> >
> > We ( currently ) don't have to consider the oracle CRS implications but
> > ( not for long ). At the moment, as you are going to do, for bringing
> > up prod db on SRDF'd side, I also have to change information via srvctl.
>
>
> Hello,
>
> About HACMP, CRS and SRDF. This are different things. HACMP is the AIX
> clustersoftware to help you in a cluster environment. CRS is the Oracle
> part doing the same.

Who didn't know they were different things?

We run hpux here where hp provides serviceguard. It functions in a similar manner to provide clustering that HACMP apparently provides.

Do you really want oracle to provide the clustering layer? Many people would like to at least have a choice of using vendor provided or taking an oracle pieces/part.

In 10g though CRS is mandatory for RAC.

>
> You can choise which to use. CRS is for "free" when building RAC. So
> just looking at the costs without HACMP it is cheaper.
>
> About SDRF. SDRF is a way to mirror your data to another side. As we
> say only the data.
> Which means you need to think off away how you will start your cluster
> environment at the other side when there is a side failure and you need
> to failover to the other location.

We are currently running SRDF. You run scripts to make volume groups available for read write access on the other side of the SRDF when you need to.

I already noted that you can use srvctl to change your RAC configuration on the other side. What part did you not read in this thread?

>
> But as mentioned only the data is at the other location, which means
> you need to think about starting hacmp or CRS onanother host, prepare
> it with the oracle software for example.

I noted early that we ( not the OP ) are currently running 9.2. No such critter as CRS there. Received on Wed Oct 04 2006 - 16:04:05 CDT

Original text of this message

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