Video Screencast Help

VVR with MSCS-clustered storage

Created: 21 Dec 2011 • Updated: 22 Dec 2011 | 3 comments
This issue has been solved. See solution.

Hi Forum,

I setup a MSCS cluster with two nodes. I use SFW 5.1 SP2 for storage virtualization. Failover clustering with the SFW virtualized volumes works just fine.

Now I want to replicate the data of the SFW volumes to another disk array and I want to use VVR for that purpose. Already installed VVR feature successfully during SFW setup. I've got two seperates disk arrays for the Primary resp. Secondary Replicator Logs.

1. I want to use the two MSCS cluster nodes also as Primary resp. Secondary Replicator Nodes. I read that you cannot add more than one single SFW disk group to a RDS/RVG, so I would have to add the Replicator Log disk to the disk group which contains the data to be replicated. As that SFW disk group is a MSCS cluster resource the RL disk would also be switched to the second MSCS node if a failover occurs. The Replicator Log disk of course would have to be connected to both the Primary and Secondary nodes.

Question1: Will the VVR system become totally confused or even destroyed, as the Secondary Node would take over the Primary Replicator Log disk in a fail over event?

Question2: Does the above described configuration work at all or do I have to setup two more Windows Servers with SFW as Primary resp. Secondary VVR nodes?

Thanks in advance!

Best Regards,

Bob 

Comments 3 CommentsJump to latest comment

Wally_Heim's picture

Hi Bob,

I'm not real clear on what you are trying to do here.  Let me see if I can answer your question by providing the following examples of what works and what doesn't work in a MSCS/WFC cluster when using VVR.

 

1.  Rplicated Data Cluster does not work with a MSCS/WFC cluster.  In this cluster, VVR would be used to replicate the shared data from disks on cluster node 1 to a different set of disks on cluster node 2.  The main reason that this does not work is that VVR support on MCSC/WFC does not have the needed resource to control VVR migration/takeover operations. 

if this type of configuration is needed, then an SFW-HA cluster using VVR would be what you need because it can handle VVR migration/takeover operations when failover between nodes in the same cluster.

 

2.  MSCS/WFC cluster using VVR to replicate data to another server outside of the cluster is valid.  AKA the cluster is the VVR primary.  In this case, the MSCS/WFC cluster would have a service group with a virtual IP resource that VVR uses to replicate data away from the cluster.  This would be a VVR primary and is a valid configuration.  Keep in mind that VVR rlinks need to be setup to use the virtual IP for replication so that it will move from node to node during failover.  The replication data set will contain all volumes beinig replicated and a unique volume for the replication log for this replication data set.

 

3. MSCS/WFC cluster using VVR to receive replicated data from outside of the cluster is also valid.  AKA the cluster is the VVR secondary.  In this case replication again is setup using a virtual IP address and the replicated data set contains all volumes that are being replicated and a unique volume for the replication log for this replication data set.  The main issue with this is that the data is not accessible since this is a VVR secondary.  A VVR migration or takeover would need to be performed via CLI or VEA before the data would be in a read/write state on this cluster.

 

4. You can setup two seperate service groups on the MSCS/WFC cluster so that you are able to replcate data from the cluster using 1 replication data set / 1 service group and receive data from outside the cluster in the other service group.  In this configuraiton each service group would have its own set of disks for volumes and replication logs and would be configured to use a unique virtual IP address for that service group.

 

Please let me know if this helps or I completely misunderstood what you are trying to do.

Thanks,

Wally

SOLUTION
unex's picture

Hi Wally,

thank you very much for your answer.

Case 1 isn't the aim.

Case 3 neither.

I think what I need is Case 2 or Case 4.

Actually it is Case 2 but I thought that I could configure one of the MSCS nodes as Primary VVR and the other node of the MSCS as Secondary VVR. Seems that I have to setup a third Windows Server with SFW and VVR and use the MSCS cluster as Primary node and the new Windows Server as Secondary VVR.

If I understood Case 4 right, I would also need a third Windows Server to act as Secondary VVR.

Thank you again! Now it's clear.

Best Regards,

Bob

Wally_Heim's picture

Hi Bob,

VVR with MSCS is intended to replicate volume(s) by being either the VVR Primary or the VVR secondary for a given replication data set (RDS).  As such, you will need either another MSCS cluster or a standalone node to be the other side of the replication data set.

Since you can have more than one replication data set replicating on a single server (standalone or MSCS cluster node) you can have on replicating data away from the MSCS cluster and the other replicating data to the MSCS cluster.  In this case the MSCS cluster would be the VVR primary for one RDS and the VVR secondary for the other RDS.  Of course it would not be for the same volumes/data.

If this answers your issue please mark the post as resolved and identify the solution.

Thanks,

Wally