VMware snapshots instead of production data.

Agree, this thread has turned into a great wealth of knowledge when it comes to snapshots. 

This is really becoming a great thread...love it!  Would be nice to have a Veeam PM chime in here…. 

a wait a storage snapshot so not as bad

wow that must have been fun. I have had really bad situations with VMs Snapshots running for weeks. 

I get that.  But in the case of a storage snapshot, I believe both the base disk and delta disk(s) would be accessible because the active workload is shifted back to the life files on the array, and not the snapped files.

Not much changes with storage snapshots

it’s just the SAN volume gets a snapshot and the VMware snpashot can be removed.

While Veeam does it’s thing, that SAN volume snapshot is a clone of the VM with the snapshot on it. 

Also good to know.  I didn’t realize that CBT didn’t apply to snapshots - I do see however that to enable CBT there shouldn’t be snapshots on a VM or that can cause errors.

I may have to do my own testing to make sure I understand how it works correctly. I had assumed that CBT included blocks on any delta disks in use as well.  But like I said, easy enough to test.

With that said, in reference to your above, it takes a snapshot in order to free up the base disk for access.  If the VM is attached to the base disk, another VM can’t attach to it to gain read access.  The snapshot offloads the access for the production VM to a delta (snapshot) disk.  This is almost sort-of difference in the case of storage snapshots.

I think the existing snapshots are transparent for Veeam. The Veeam Snapshot create a fixed state of the VM including the already existing snapshots. So, the backups should be fine.

Although, the existing snapshots prohibit the usage of changed block tracking (CBT). Therefor the speed of the backup will be slower than without additional snapshots.

And the snapshots will fill up the storage of the cluster...

There is no data loss backing up a VM with a snapshot. You can have many snapshots on a VM and when you run the backup you get the CURRENT stat at the time the backup was run. 

I am testing this as I write it. 

1. Creating a file on desktop calling it BEFORE_SNAPSHOT

2. Take a VMware Snapshot

3. Renamed the file to AFTER_SNAPSHOT
4. Run a Veeam backup job containing VM
5. Do a file level restore for the VM in Veeam
6. Navigate to C:\Users\XXXXX\Desktop

File shows AFTER snapshot.  (as expected)

I am using storage snapshots but this is the order.

Veeam tells VMware to take a snap, the SAN does a volume snapshot, then VMware deletes the local snapshot and it retrieves it from the SAN.

No stun and lightning fast. (other than initializing the job)

Think about it., If Veeam doesn’t backup the “last” snapshot, why would it be taking a snapshot at all.

The terminology is a bit odd, It’s true Veeam doesn’t backup VMware snapshots. . You could have 8 Snapshots on a VM, run Veeam, and you will not backup those 8 snapshots. What you will get is the CURRENT state of the VM the time the backup was run.  You can’t backup a VM “with” the snapshots. As in you can NOT restore a VM and have those 8 snapshots attached. It will only be the state of the VM when the backup was run.

Hope this helps.

I just consolidated a 140TB Snapshot :) 

Sure thing @dips 

Thank you @coolsport00 Massively helpful :) 

The Snapshot Hunter feature Veeam created several years ago was much needed. Mostly takes care of orphaned snaps. You are correct though...if, for whatever reason, Veeam can't /doesn't successfully remove the snap after a backup, it can grow. It's highly advised to never manually delete them or yes...you will lose data. Veeam does continue trying to consolidate using the Hunter. If it still can't remove it, all that's needed is to create a snap manually, let it go for like a minute, then consolidate them all. 

So yeah, those snapshots are there to store the changing data in a delta. If you delete them you lose the data that changed since the start of the backup. I guess when they get orphaned they manage to consolidate or something hence no issues deleting them? I mean otherwise if no one checked that thing could grow to be a monster and then you might not be able to consolidate. 

@Geoff Burke no worries...good reminder for myself as well

ah good call @coolsport00 ! Correct it does state that clearly here “The source Veeam Data Mover reads the VM data from the read-only VM disk and transfers the data to the backup repository in one of transport modes.”

https://helpcenter.veeam.com/docs/backup/vsphere/backup_hiw.html?ver=120

@dips - no sir. The screenshot below is what Veeam says in their documentation:
 

You can read more here.

Is the behaviour for taking backups from Storage Snapshots the same @coolsport00 

No worries @dips ...it was a good reminder for myself as well 

Thanks @coolsport00 I definitely learnt something new today! I’ve always thought it was the snapshot that got backed up rather than the VM but  that makes total sense. 

Yikes….good to know.

@dloseke  Veeam doesn’t back up any snaps. If you have 6 snaps on a VM (user-created, for ex.), they don’t get backed up...just the point-in-time of the parent disk before the 1st snap was taken.

Shane is right on this one for sure.