With the release of Purity//FA 6.2.6 and Everpure vSphere Remote Plugin 5.1.0, two new features of VM Undelete were released: the ability to revert a VM to a specific Point in Time (PiT) array based snapshot and being able to Undelete a vVols based VM that has been eradicated on the array to a specific PiT array snapshot. The standard VM Undelete is still present.
Before getting into the workflows specifically, let's cover what is required in order to be able to execute the PiT Revert and PiT Undelete workflows.
- In order to be able to execute VM Undelete within the eradication timer (default 24 hours), at minimum an array snapshot of the VM's config vVol is required.
- In order to be able to execute PiT VM Undelete of a vVols based VM that has been eradicated, a FA protection group snapshot of all the VMs Data vVols, managed snapshots and Config vVols are required.
- In order to be able to execute a PiT VM Revert of a vVols based VM, a FA protection group snapshot of all vVol disks of the currently configured VM are required.
- The volume snapshot objects should be associated with an FA protection group directly and not through a host object or host group attached to the protection group.
- If the ability to consistently back up the environment with snapshots is the goal, when the VM is powered off while a snapshot is being taken of the host or host group object, but then gets powered on after the snapshot interval, the protection required might not be present because that VM would not have had a snapshot taken of it. This applies if the VM is in-flight during the snapshot operation as well.
- The vVol VM can shift hosts with DRS enabled in vCenter or can be manually migrated to other hosts; this changes where the volumes are mapped from a FlashArray host object perspective.
- The host will disconnect from some of the vVol volumes when the VM is powered off and while the VM is powered on, the swap volume would be backed up which is unnecessary.
- The vVols service on the FlashArray, VASA, won't have insight into why those volumes are part of the protection group. Depending on the SPBM policy or replication group applied to the VM, there might be misleading compliance results on the VM in vCenter.
- The volume snapshot objects should be associated with an FA protection group directly and not through a host object or host group attached to the protection group.
With that covered, Everpure does recommend leveraging SPBM and local snapshot protection placement rules for vVols based VMs. Please see the FlashArray Storage Capabilities for more information on local snapshot protection placement in particular.