How To Map An RDM To Multiple VMs

Sometimes you need to map an RDM to multiple VMs in VMware, if you are creating a cluster for example. You will probably find that when you go to map the RDM to each VM after the first, the option to select RDM is greyed out. This is because you can’t map an RDM to multiple VMs. To get around this and map an RDM to multiple VMs:

  1. Add a new disk to the first VM you want to add an RDM to (Right click the VM -> Edit Settings… -> Add… -> Hard Disk -> Next -> Raw Device Mappings, etc…)
  2. Next, add a hard disk to the other VMs that you want to map the RDM to, except this time, specify that you want to add an existing disk, navigating to the vmdk file that will be created on the datastore containing the RDM that you mapped in step 1

Clustering Windows Servers in VMware

If you are mapping an RDM to multiple VMs for the purpose of creating a cluster in Windows, make sure to add the RDM to a new SCSI controller (other than the one used for the OS disks) and to map each RDM to a new SCSI ID on the other controller. Note that you also need to use a compatible SCSI controller (that’s identical across the multiple VMs in your cluster) 😉

VN:F [1.9.22_1171]
Rating: 0.0/10 (0 votes cast)

2 thoughts on “How To Map An RDM To Multiple VMs

  1. I’ve got a cluster setup in this configuration. The only annoying thing is that you can’t snapshot the vm. Do you have any workarounds?

    VA:F [1.9.22_1171]
    Rating: 0 (from 0 votes)
    • Unfortunately I have no idea how to snapshot a VM with an RDM. I’m actually going through this same issue myself, it’s a shame to go through the hassle of working out how to map an RDM to multiple VMs only to find that you can’t snapshot the VMs 🙁

      My solution was to question whether I really need to use a Microsoft cluster, considering that VMware is already providing HA / FT. I suppose that a cluster does protect the Windows layer, rather than just the hardware layer, but how many times have you seen a well maintained Windows VM randomly get corrupted? More likely to get damaged after some administrative action on all nodes in the cluster, such as performing Windows Updates, in which case clustering won’t save you anyway. My solution was to just use a normal VMware datastore, rather than RDM and just have a single VM rather than a cluster. Better to do it this way and have an improved RTO with a snapshot based backup solution and enjoy less complexity of managing a clusterless VM without an RDM.

      The only place I can see a cluster actually being required is in an active-active setup, for performance in large environments… In which case, you probably have more dedicated resources and a large amount of data… In which case, you are probably doing SAN level snapshots instead 😉

      VN:F [1.9.22_1171]
      Rating: 0 (from 0 votes)

Leave a comment

Your email address will not be published. Required fields are marked *