Thin Provisioning

Thin-provisioned virtual disks on vSphere enable you to overcommit space at the datastore level. Overcommitting space means that you can provision more space to VMs than the actual capacity of the datastore allows. If the environment has a large number of VM deployments with unpredictable space growth, using thin-provisioned virtual disks on thin-provisioned VMFS volumes might be a viable solution. However, you must exercise caution to prevent space overcommitment at the VMFS datastore level from causing out-of-space conditions for VMs.

If you do not use thin-provisioned virtual disks, ensure that the Datastore usage on disk alarm is enabled at the highest level of the vSphere infrastructure. Also make sure that the alarm has an acceptable value range for the warning and alert triggers. In vSphere, the predefined default value for warnings is 75% of disk space usage; the value for alerts is 85%.

When you receive a warning or alert, you can use the HPE Nimble Storage vCenter plugin to resize datastores. In addition, in a VMFS datastore that contains both thin-provisioned and thick-provisioned virtual disks, you can correct the out-of-space condition by converting thick-provisioned virtual disks to a thin format. You can convert them through a storage vMotion migration to another datastore, followed by a migration back to the source datastore.