What is a checkpoint?
A checkpoint is a saved state of a Virtual Machine at a point in time that can be returned to. Other virtual solutions (notably VMWare) refer to this as a "snapshot." Multiple checkpoints can be created on a single VM, enabling one to jump between machine states quickly and easily for testing purposes.
Creating a checkpoint creates a journal file on the datastore, that logs all of the disk write operations. The original hard disk is unmodified.
Some third party backup software hooks into Hyper-V and creates checkpoints to use during a backup. It is important that any software that does this cleans up after itself and removes old checkpoints.
Is a checkpoint the same as a backup?
No. A checkpoint is merely a saved point in time that can be returned to. It is in no way a backup and does not provide any data security.
Important: It is critical to use the correct terminology when discussing snapshots and backups. Misunderstandings can lead to data loss!
Advantages of a Checkpoint
- Quickly create a point in time that can be easily returned to
- Keep several different configurations ready to go for testing purposes
Drawbacks of Using Checkpoints
- Running a VM with an active checkpoint has a performance cost due to writing to the the journal file mentioned above. More checkpoints = more performance cost.
- Creating too many checkpoints can be confusing to administer.
- Leaving a checkpoint active will eventually fill up the disk that the VM resides on. The Journal file grows with each disk write operation, including deletes.
When should I use checkpoints?
Normally you should run your system with no checkpoints enabled. This will provide the best performance.
You should create a checkpoint before any major configuration changes, including Kodak Workflow Software upgrades. This will enable you to very quickly return to a working state should something fail during the upgrade.
Note: Please power down the VMs before creating a checkpoint. This is much faster an guarantees application and database consistency.
You should delete the checkpoint after a suitable amount of time has passed and you are satisfied that the upgrade worked properly (usually a maximum of 1 week).
You do not have to power off the VMs when deleting a checkpoint. Performance will be a bit slow while the journal file is being reintegrates with the virtual hard disks, but will return to normal once the deletion is finished.