Tutorial: How to Move from One Storage Platform to Another with MS Cluster Servers & RDMs in VMware

By Brian Singh
published: Thursday, March 26 2009

Today, the movement of Virtual Machines between hardware hosts is a fairly simple and seamless operation. Migration using VMware's ESX 3.5 and Virtual Center 2.5 employs the VMotion process to migrate each VM from one ESX host to another with, at most, a few clicks of the mouse. Storage VMotion, announced December 2007, further enhances portability and migration efforts by facilitating the movement of the actual vmdk (virtual hard drive) files to other, alternate storage locations without downtime. It's all quite simple, right? Well, yes...until we encounter Raw Device Mappings. RDMs, in short, are a special file on a VMFS volume that serves as a proxy for a VM to access a raw device in a direct manner. Unlike a VMFS partition, VMware's ESX Server does not manage the device when it is in physical mode. It is up to the virtual machine's operating system to manage it directly.

For functionality, applications like Microsoft Cluster Services depend on the shared storage presented in the raw, or RDM, format. For those looking to switch storage arrays or platforms or entirely migrate these VMs from one ESX cluster to another, this creates a few, albeit surmountable, challenges and a bit of additional planning to ensure the integrity of cluster. To get there from here, new RDMs must be created and configured to be identical to the original RDMs and that's exactly what we will show you how to do in the rest of this article. Although this article comes with no guarantee, rest easy, the procedure presented to you preserves the original VMs and RDM configuration in the event that the dreaded rollback is desired.

For our procedure we will be using:

• ESX 3.5 update2
• Virtual Center 2.5
• Windows 2003 Enterprise Ed.
• Microsoft Cluster Server
• Windows 2000 resource kit

The process is broken down into 8 basic steps. Each step has a corresponding "Section" in the tutorial for quick reference. A word on typographic convention, all text set off by single quotes and bold (‘foo command') refers to a typed system command and the associated parameters and arguments. Prompts such as Click, Select or Launch are capitalized while inline notes are prefaced with Note -  and in italic 8 point Arial text:

1. Record the Original Configuration
2. Add New LUNs
3. Add the RDM Configuration
4. Copy Data
5. Clone the VM
6. Configure the Clone
7. Change the Microsoft Disk Signature
8. Add RDMs to Additional Nodes in the Cluster

Step 1 - Record the Original Configuration

1.  Login to Virtual Center right click on the Primary Node within the Cluster configuration. This server will be referred to as Node1.

2.  Select, Edit Setting.

3.  Document all the SCSI controller settings.

4.  Document each Raw Mapped LUN, Datastore Mapping and Virtual Device Node.

5.  Login to Node 1.

6.  Click, Start, Click, Run, type ‘compmgmt.msc', and then Click, OK.

7.  In Computer Management under System Tools, Device Manager, look under Disk Drives, and you can view all the logical disks that are being presented to the node.

8.  Document the disk number on the left side and drive letter associated to it on the right side. (Example: Disk 1 =  1GB Quorum Q drive, Disk 2 = 10GB Data  E drive)

9.  Copy, dumpcfg.exe (windows 2000 resource kit) to c:\.

10.  Click, Start, Click Run, Type ‘cmd', Click OK.

11.  Type ‘cd to c:'

12.  Run ‘dumpcfg'.

13.  Document the disk signatures with the disk numbers recorded above.

      (Example: Disk 1 = Quorum Q drive =1234567: Disk 2 = Data E Drive = 7654321)