HP StorageWorks SAN Virtualization Services Platform 2.1.1 release notes (5697-0066, July 2009)

• Do not mix virtual disks with both “Continue-on-fail” and “Always sync” synchronous mirror groups
in the same cluster.
Operational best practices
• Should you want to delete, remove, or undo an operation to an SVSP entity, it is best if you reverse
the processes that were used to get you to that operational stage. Additional details are in the HP
StorageWorks SAN Virtualization Services Platform administrator guide.
For example, you have I/O to a snapshot created from a PiT that was created from a virtual disk
(that is online to a different application). Assume the following steps were performed to get to this
stage:
1. A virtual disk is created from a pool.
2. The virtual disk is presented to a host.
3. The first I/O stream is initiated to the virtual disk.
4. A PiT is created from the virtual disk.
5. A snapshot is created from the PiT.
6. The snapshot is presented to a host.
7. The second I/O stream is initiated to the snapshot.
If you decide to delete the virtual disk, you should remove or halt all the operations that were
performed from the time the virtual disk was created, by working backwards. In this case the
process would be:
1. Halt the second I/O stream to the snapshot (initiated in step 7 above).
2. Unpresent the snapshot to its host (presented in step 6 above).
3. Delete the snapshot (created in step 5 above).
4. Delete the PiT (created in step 4 above).
5. Halt the first I/O stream to the virtual disk (initiated in step 3 above).
6. Unpresent the virtual disk to its host (presented in step 2 above).
7. Delete the virtual disk (created in step 1 above).
Carrying this example further, if you decide to delete the pool from which virtual disks are created,
you must delete all the virtual disks in that pool. This requires that you delete or halt all operations
being performed on or by those virtual disks, like I/O, snapshots, PiTs, presentations, and so on.
• HP recommends that you limit the number of virtual disks in a pool to 500, with a maximum limit
of 1000. In addition, avoid creating virtual disks that span more than one pool.
• Do not remove a single snapshot from a virtual disk group snapshot set.
• When a threshold limit is reached, a single event log message is generated, instead of a message
being sent every five minutes.
• Restoring a VSM setup from a backup can be safely performed only if there are no PiTs in the
system. If PiTs exist (either created directly or by Business Copy or Continuous Access applications),
their metadata in the setup backup may be invalid and may result in data loss/corruption if restored.
• Do not change any back-end logical unit characteristic (for example, number, name, size [by ex-
panding or contracting], and so on) once it has been virtualized (for example, entered into a VSM
storage pool) or is being managed by the VSM. The LUN identification has been written on the
LU by VSM.
8