You can use datastore clusters in the ESXi clusters that host VMware Integrated OpenStack compute workloads.

A datastore cluster is a collection of datastores with shared resources and a shared management interface. You can use vSphere Storage DRS to manage the resources in a datastore cluster. For information about creating and configuring datastore clusters, see Creating a Datastore Cluster in vSphere Resource Management.

If you want to use datastore clusters with VMware Integrated OpenStack, be aware of the following:

  • Datastore clusters cannot be configured by using the Integrated OpenStack Manager web interface. To add a datastore cluster to your deployment, perform the following steps using the command-line interface.
    1. Specify the name of the Nova Compute server to update. 
      viocli update novacompute <novacompute_XXX>
    2. Ensure that your datastore cluster is accessible to the Nova Compute server.
    3. Assign the name of your datastore cluster to the datastore_cluster key.
      conf:
  nova_compute:
    DEFAULT:
      default_schedule_zone: nova
      disk_allocation_ratio: 2
      host: compute01
    vmware:
      cluster_name: compute_cluster
      datastore_cluster: <your_datastore_cluster>
    4. Wait for the Nova Compute server to restart.
    When you use Nova to start a new compute instance, it is allocated to your datastore cluster.
  • Only one datastore cluster can be used for each vCenter Server instance.
  • If your environment has multiple vCenter Server instances, the name of the datastore cluster used by VMware Integrated OpenStack in each instance must be the same.
  • Swift nodes do not support datastore clusters.
  • You can only boot images backed by virtual machines. Sparse and preallocated images cannot be booted on datastore clusters.
  • You must enable Storage DRS on your datastore clusters and set the Cluster automation level to No Automation (Manual Mode). Automatic migrations are not supported.
  • Only the following provisioning operations use Storage DRS:
    • Booting from a Glance template image
    • Creating raw Cinder volumes
    • Creating a volume from another volume (full clones and linked clones)
    • Cloning snapshots in COW format (full clones and linked clones)