Storage options for a Windows failover cluster on vCenter 5.5

[AlphaAdmin]

Hello, I am running vCenter 5.5 Essentials Plus with 3 VMHosts and an attached SAN. I want to create a Windows Failover Cluster (WSFC) that will be used for a SQL cluster. This is my first attempt with WSFC so please excuse the newbie questions.

I understand that the cluster will need shared storage. I see two options:

1. Create an additional partition in the SAN, give it another LUN, make the partition available to the hosts, then use it as a "data" partition in my cluster VMs

2. Create a virtual disk (.vhdx format) in the existing SAN partition and share this between the cluster VMs

Questions:

- My initial thought would be to go with No.1, this way I eliminate the "middle person" and connect directly to the LUNs. Would this be a good approach? What are the pros/cons between the two options?

- Is option 1 even possible? I created the additional partition on the SAN and I cannot figure out how to "see" it from the hosts. Do I need dedicated physical connection for each different LUN I want to connect to?

- Are there other options that would be preferable in this case?

Much thanks! 🙂

Christos

[vThinkBeyondVM]

You may want to refer this white paper:http://pubs.vmware.com/vsphere-55/topic/com.vmware.ICbase/PDF/vsphere-esxi-vcenter-server-55-setup-m...

Also refer:VMware KB: Microsoft Cluster Service (MSCS) support on ESXi/ESX

[JPM300]

Hey AlphaAdmin

You will need to decide how you want your Microsoft Clusters Shared Data drives to work.  These drives will be your Quorum drive and your Data drive for your SQL.  These will need to either be a RDM in virtual mode or a RDM in Physical Mode.  A Virtual RDM gives you more control over VMware abilities like vmoition ect.  However both Microsoft Cluster nodes need to stay on the same ESXi host.  If you make the RMD physical they can be on seperate ESXi hosts however you can no longer vmotion.  HA will still work with either so if you go with physical you will just have to flip Microsoft cluster nodes whenever you want to do ESXi maintenance. 

If you have any questions please let me know,

[AlphaAdmin]

Hello and thanks for the reply. Naturally I now have more questions! 🙂

My original post talk about .vhdx virtual disks .vs. LUN. It seems to me that both options you presented refer to the latter. Is this correct?

In the .vhdx scenario, do I need a separate SAN partition with a separate LUN? My understanding is that .vhdx is just a shared virtual disk in the VMWare infrastructure.

In the RDM scenarios, do I need a separate SAN partition and where/how would I connect it in each case? Do I need a separate physical connection (an actual physical SAS cable) if I want to "see" the additional partition from the VMware infrastructure or can I "see" it through the existing SAN connections/cables?

Thanks

Christos

[JPM300]

Hey Alpha,

Sorry for the long reply

If you do a shared vmdk it is just a shared virtual disk living in the same Datastore.  However if you do a RDM this is shared out to any ESXi hosts you would want the MSCS Nodes to be able to boot up on with HA.  It would also be shared to both MSCS cluster nodes.  However there is a step by step process on this

Here is a quick video on it.  Don't worry about the language or the bad audio, it goes through the steps in the gui to get you squared away

https://www.youtube.com/watch?v=q_871LHjgGM

Hope this has helped

[AlphaAdmin]

Hello and sorry for the long silence. I got a lot of information but I am still unclear about a basic design question: Is it possible to have the nodes on separate physical hosts (no cluster-in-a-box) AND have VMware HA if a host dies (the VM will restart on another host)? If yes, do I need to use Pass-Through RDM (physical compatibility mode) or Non-pass-through RDM (virtual compatibility mode)? Thanks.