Replicated LUNs and independent LUNs within a DSS Cluster

[cneri]

Hi,

Is it possible in a Replicated scenario with failover and MPIO (with ESXi 4.1 in iSCSI)

(like this : http://documents.open-e.com/Open-E_D..._XenServer.pdf)

to have both Replicated LUNs et non-replicating LUNs (independant on each array).

May i have to configure 2 VMkernels by card on my ESXi to be able to contact both direct array IPs and Virtual IPs ?

Is there a better way to do this ?

I'm planning also to setup a 4 paths MPIO network. Is there any recommandation with this ?

Thanks

[Al-S]

For MPIO, please find the more info at our webcaset and videos:
http://www.open-e.com/library/webcasts-and-videos/
Open-E DSS V6 MPIO with VMware ESX4

[cneri]

Ok thanks, i've already saw this video but the demo is a 2 paths MPIO. My question is, is there any additionals things to take care with 4 paths ?

But the main question of the post is about combining replicated LUNs and independant LUNs in a 2 nodes DSS cluster.

Thanks

[To-M]

We have only tested with 2 NICs but this would depend on the initiator side to add more NICs for the MPIO.

For the Auto Failover the Primary server will need to set the volumes in Source mode and the Secondary server set them in Destination mode and you cannot mix them if using the Auto Failover. Though you can use it in such a way to mix them just using the Volume replication function only.

[cneri]

Ok thanks and in the case of using different Volumes instead of different Luns wil that do the trick ?
(1 volume source/destination, 1 volume independant).

[To-M]

The iSCSI Volumes dont have to be in used for Replication so they can be accessed is not assigned with replication.

[cneri]

Ok and last thing about the auto-failover limitation in this case (because it's becoming a little bit complicated for my understanding ).
I dont understand why i cant attach my clustered volumes with the failover virtual IP and in same time attach independant volumes (non redundant) using the physical array IPs ?

The aim is to keep some storage space seperated from synchronous tasks but keeping autofailover on clustered volumes would be great. Don't you think that would be a valuable function ?

Best Regards

[To-M]

But you can use your clustered servers like VMware's HA or MS Cluster we have videos showing this from link below and many use the DSS V6 for this setup. If you are having issues connecting from your iSCSI initiators this link below might provide some help.

http://www.open-e.com/library/webcasts-and-videos/

http://www.open-e.com/library/how-to-resources/