Exchange 2003 Standby Cluster

Thread view:

Enable EMail Alerts

Start New Thread

Thread rating:

AndyJ - 28 Dec 2006 09:06 GMT

Hello All

I am planning for a single node standby cluster at our DR location so
we can perform a tape restore of our production cluster in the event of
a disaster and just wanted to understand the requirements regarding the
disk configuration. I am proposing that we perform full computer
backups of our production cluster nodes.Can someone tell me if the disk
partitions on the standby single cluster node and the drive letters etc
have to be the same. The production cluster will be SAN attached and
the single node cluster will just have direct attached storage. Other
than that the hardware will be identical.

>From what I read the process is pretty simple:

First restore the full computer backup of the node that owned the
cluster - which will restore the quorom and then restore the Exchange
databases. Its a single node cluster so will not have any other nodes.

Anything obvious that I am missing here?

Thanks

AJ

Reply to this Message

AndyJ - 29 Dec 2006 10:04 GMT

That was a typo BTW I meant the node that owned the Qurom not the
cluster!!!

Also in addition to the below. I am planning on just having the standby
cluster installed with Windows 2003, in a work group with a randomly
generated machine name so I can just blast the full computer backup
over the top.

Is this approach sound and valid?

Thanks

AJ

> Hello All
>
[quoted text clipped - 19 lines]
>
> AJ

Reply to this Message

Leif Pedersen [MVP] - 29 Dec 2006 14:31 GMT

Hi,

This won't work since the cluster service account needs to be a domain
account.

Leif

> That was a typo BTW I meant the node that owned the Qurom not the
> cluster!!!
[quoted text clipped - 33 lines]
>>
>> AJ

Reply to this Message

AndyJ - 29 Dec 2006 14:46 GMT

Hi Leif

I am not sure I am following you here could you elaborate?

Perhaps it was my wording. I mean that the standby server (Non
clustered at this point) would just be running Windows 2003 Enterprise
so I could restore the full computer backup to it.

Thanks

AJ

> Hi,
>
[quoted text clipped - 40 lines]
> >>
> >> AJ

Reply to this Message

Russ Kaufmann [MVP] - 30 Dec 2006 00:22 GMT

> Hi Leif
>
[quoted text clipped - 3 lines]
> clustered at this point) would just be running Windows 2003 Enterprise
> so I could restore the full computer backup to it.

And that is not the way stand by clustering works.

Signature

Russ Kaufmann
MVP - Windows Server - Clustering
ClusterHelp.com, a Microsoft Certified Gold Partner
Web http://www.clusterhelp.com
Blog http://msmvps.com/clusterhelp

The next ClusterHelp classes are:
Denver starting Feb 12th
NYC starting Feb 19th

Reply to this Message

AndyJ - 30 Dec 2006 17:33 GMT

Ok Russ, so I guess I have my terminology wrong.

I want to have an Exchange server that will be at a DR location that
can be used to restore my Exchange cluster too in the event of a
disaster at both of our data centres. We already have NSI Double Take
replicating from the primary cluster to a single node cluster in our
second data centre but management have made it clear to me that they
want, in addition to this, a server that can be used to restore the
production environment to if they lose both data centres at a third
location. This restoration process will simply be from tape at the Dr
location so no restores over the WAN etc etc.

Now what I described is that not what I need to do?

I am not intersted in failing over an EVS to a standby cluster. The
server in the DR location is purely there for restoring the clustered
servers stores too.

Thanks

AJ

> > Hi Leif
> >
[quoted text clipped - 16 lines]
> Denver starting Feb 12th
> NYC starting Feb 19th

Reply to this Message

Russ Kaufmann [MVP] - 31 Dec 2006 00:44 GMT

> Ok Russ, so I guess I have my terminology wrong.
>
[quoted text clipped - 13 lines]
> server in the DR location is purely there for restoring the clustered
> servers stores too.

Why not? Using a standby cluster (which can be a single node) will enable
you to recover a cluster faster than most other options. Please read the
article:
http://www.microsoft.com/technet/prodtechnol/exchange/Guides/DROpsGuide/d2492381
-e9d5-4827-b7ff-7ac614ccc36d.mspx?mfr=true

The process is incredibly easy and will get your organization back online
quickly without having to touch your client machines.

Signature

Reply to this Message

AndyJ - 31 Dec 2006 08:55 GMT

Good point, why not! :) and it was staring at me right in the
face!!!!!! I wrongly thought that a standby cluster could not be used
in a single node configuration.

So in a disaster situation at a high level they would failover the EVS
to the standby cluster and then simply restore the databases from the
tape backup

On reading the article that would be a much better strategy

Thanks
AJ

> > Ok Russ, so I guess I have my terminology wrong.
> >
[quoted text clipped - 32 lines]
> Denver starting Feb 12th
> NYC starting Feb 19th

Reply to this Message

Russ Kaufmann [MVP] - 31 Dec 2006 10:45 GMT

> Good point, why not! :) and it was staring at me right in the
> face!!!!!! I wrongly thought that a standby cluster could not be used
[quoted text clipped - 3 lines]
> to the standby cluster and then simply restore the databases from the
> tape backup

Again, a point of terminology. You don't failover to the standby cluster.
You re-create the EVS on it. The EVS has the same name so clients can
continue to access it by name. The toughest part of it is making sure to
delete and recreate the EVS computer account in AD and updating the IP
address in DNS.

I would suggest building a test cluster in Virtual Server 2005 and then
creating another (standby) in VS 2005 and testing it out.

Signature

Reply to this Message

Russ Kaufmann [MVP] - 30 Dec 2006 00:21 GMT

> That was a typo BTW I meant the node that owned the Qurom not the
> cluster!!!
[quoted text clipped - 3 lines]
> generated machine name so I can just blast the full computer backup
> over the top.

Again, you are missing what standby clustering is all about. You should
build the single node cluster BEFORE the failures of Exchange and thus it
must be part of the domain and use a domain account for the cluster service
account.

Signature

Reply to this Message

Russ Kaufmann [MVP] - 30 Dec 2006 00:20 GMT

> I am planning for a single node standby cluster at our DR location so
> we can perform a tape restore of our production cluster in the event of
[quoted text clipped - 3 lines]
> partitions on the standby single cluster node and the drive letters etc
> have to be the same.

The drive letters must be the same, but the sizes do not have to be the
same.

> The production cluster will be SAN attached and
> the single node cluster will just have direct attached storage.

No can do. The disk must appear as cluster-capable.

>>From what I read the process is pretty simple:
>
> First restore the full computer backup of the node that owned the
> cluster - which will restore the quorom and then restore the Exchange
> databases. Its a single node cluster so will not have any other nodes.

I don't know where you are reading this, but that is not what standby
clustering is all about.

The standby node does not have to have the same name as any of the
production nodes. The EVS name is what it is all about. Please read the
documentation again on standby clustering.

Signature

Reply to this Message