Multiple Sites

[Kris Wolton]

Hi,

We have 2 P4300 clusters setup for a customer over two sites connected by a 10mb/s WAN link. We had to set them up kinda quickly as the previous storage platform died. We followed an HP/Vmware best practice guide but somehow managed to get it wrong.

Basically at the moment we have both 2 sites in the same cluster (seemed logical at the time) but the problem we have is everytime the primary site goes offline (looses comms, power, etc) the DR site also goes down, which is less then useless.

Apparently we need to setup a third site to be a whitness, then delete the 2nd site and create it in a brand new cluster. Is this right?

I can't imagine many people have a third site laying around with reliable comms they can just load some whitness server software on?

Basically what we are trying to achieve it 2 separate sites (no need for automated failover as this is done by the software/vmware) with the ability to copy snapshots to the DR site.

Please let me know as i have spoken to several people about this and had several different answers.

[Bart_Heungens]

Hi,

From what I can see on the picture is that U created 2 clusters...

From what I understand U need 1 cluster with 2 sites underneath... And yes a third site which hosts the so called FOM that will be the arbitrator...

Have done several setups like that, works perfect...

Kr,

Bart

[oikjn]

The basic problem is that you have four nodes and are using a virtual manager instead of a FOM.

I don't totally get your setup plan here since you have setup two clusters in the same management group that are on two different sites.  The only reason I would put nodes from different sites into the same management group would be if I was doing a multi-site cluster using syncronous data instead of what I assume is snapshot replication in your case.

In your case it looks like you are using either snapshot replication or maybe software raid on the host (don't recomend that).  Assuming you are using snapshots, the best thing I would suggest is that you move the DR site out of the management group and into its own group.  You can still do the same snapshots, but then the two clusters are 100% independent.

Either way, you have an even number of nodes, so you really need to use a FOM and not a virtual manager.  A virtual manager resides on a single node which basically gives that one node two votes...  What that means is that in a two node group, if you lose the node with the VM, you lose quorum anyway and all the VM would protect against would be the loss of the other node.

[Joris Vliegen]

Hi Kris,

The thing you want to do is called asynchronous replication. Then you don't need a failover manager on a third site. You would need to make two management groups and copy snapshots from management group 1 to management group 2. If you would want do do it the way things should be done you would need a Fail over manager on each site.

You can't copy snapshots between two clusters. Site recovery manager will be able to bring the snapshot online on the second management group.

Kind regards,

Joris Vliegen

______________________________________________

If my post was useful, click on my KUDOS, "White Star".

[Jitun]

As you have mentioned
"two sites connected by a 10mb/s WAN link."

You most probably will not/should not be a MultiSite Cluster.

You need a DR Site.

So 2 Management Groups with Remote Copy Schedules for the Volumes.

One thing to remember is, you will have 2 Management Groups with 2 Nodes each.
I would strongly recommend Adding Fail Over Manager on both Management Groups.