Impact of converting to Thin disks

[pkelly_sts]

We're planning to migrate from storage that supports thin provisioning (so we originally provisioned thick VMs onto thin storage) onto storage that doesn't so we're thinking about converting the thick VMs to thin on the way. I'm aware that, at the very least, sVmotion tends to temporarily break CBT causing longer but not larger backups.

What is the impact of adding in thick-to-thin conversion to an sVmotion? Will it cause a large backup due to changed blocks, or is it true that they're not actually changed blocks, they're just the same blocks in different places?

This is within a solution that does local backups, backup copy job to DR site as well as replication to DR site (replication jobs at default, no changes so far to vm disk configs on replication jobs).

Thanks in advance

Paul

(I did a quick forum search but couldn't see anything quite the same as I'm asking?)

[dellock6]

Hi Paul,
Veeam has the ability to identify changed blocks and also unused blocks, and in both cases to skip them during a backup operation. It can use CBT if it's available, or otherwise leverages its own capability to scan vmdks and identify changed/unused blocks. So moving a vmdk from thick to thin in my opinion is not going to change the final size of your backups. Maybe they could take longer if CBT breaks for sometimes after svmotion, because Veeam's own vmdk scan needs to read the whole disk, while CBT offers a quick map of changed blocks.

Luca.

[pkelly_sts]

Thanks for the reply Luca.

I can live with the slow-down, what I'm hoping to avoid is either a huge incremental, or needing to re-seed a load of replicas. What you say makes sense but sometimes sensible doesn't happen in reality

I think a good approach would be to do a test with one of the smaller VMs & see what happens.

Paul.

[dellock6]

Oh, a Proof of Concept is always the best way to check theory. Please let me know if I said something stupid or not

Luca.

[larry]

Not a veeam thing but on your topic. Be sure other admins don’t think the new thin disks a free. I had a SQL admin see that he now had 2tb free on each drive. He went and made copies of some databases causing the LUN to run out of space so I couldn’t remove the VM snapshots. Then my backups took much longer because of the changes. The more space you give them the more they use.

[pkelly_sts]

Good advice Larry, I can see how apps guys could get excited at the sight of a load of new "free space" In our case it's a small team that knows to ask before doing anything significant so we have no worries here.

As for our test, I converted a small (30Gb) VM to thin yesterday (and sVmotnioned it back to the original VMFS again) and can't say I noticed *any* difference to backup at all - 540Mb read, 110Mb transferred, in 2.5 mins for this VM which is in line with all the other VMs in the backup set.

So, at first test it looks like I shouldn't have much to worry about!

Thanks again Luca...

[dellock6]

No problem Paul, thanks to you for confirmation. I'm not stupid (yet)

[pkelly_sts]

As a final confirmation in case it's useful to anyone, I can say that I completed all the thick->thin migrations and it didn't impact Veeam at all, great job!

[dellock6]

Great, and thanks for coming back to update the thread

Luca.