Re: Data Centre Failover and Oracle Grid

So far testing for this system has been done on a 2 node cluster in a machine
room on site where I work, with a crossover cable between the 2 nodes The production system is proposed to be set of nodes in a single data centre with a DR in the second data centre.

I am reviewing the possibility of a cluster over both data centres, I.e. a 4 node cluster with 2 nodes in each data centre. Which created my question about the SAN's in each data centre.

Your question regarding the interconnect is a little beyond me. I am very new to 10g and clustered systems. All I can say is that the connection between each data centre is a 10Gb dark fibre, and the distance between sites is about 60 km.

I have been searching the web to try and get more of an understanding about interconnect.
I have seen within this news group a thread on interconnect protocols udp / tcpip. My networking skills are, to be honest, poor. If there is a link you could suggest, or a recommendation I would appreciate it.

"DA Morgan" <damorgan_at_x.washington.edu> wrote in message news:1114463366.100863_at_yasure...
> Alex White wrote:
>
> > Hi
> >
> > Has anyone got any experience of deploying a 10g cluster over
> > multiple data centre's. I am investigating the feasibility of deploying
> > one over two Data Centres.
> >
> > My biggest issue is with regards to the SAN based in each data centre.
> > Replication etc. The SANS are HDS and I am not sure how replication can
be
> > achieved, either bi-directional, or single san access and something like
HDS
> > truecopy
> > replicating with the second data centre.
> >
> > The SAN's are 150 terabytes in size and are predicted to grow at the
same
> > size year on year for 10 years.
> >
> > Oracle is very small but the storage is required for external objects.
> >
> > Any help would be useful.
> >
> > Alex
>
> I find your request confusing. Are you asking about separate clusters
> at different data centers replicating between the clusters or do you
> think you are going to have a cluster with different nodes at different
> locations? There is a huge difference.
>
> When answering please describe the methodology you intend to use for
> the mem interconnect.
> --
> Daniel A. Morgan
> University of Washington
> damorgan_at_x.washington.edu
> (replace 'x' with 'u' to respond)
Received on Tue Apr 26 2005 - 07:03:14 CDT