Backup Time Increases After SnapShot Delete

[jholton@nmfa.net]

Hi,

Just wondering if anyone else is seeing this behavior:
VEEAM - 6.5.0.128
VSphere - 5.0

Backup Type - Reverse Incremental
Example of 1 Server

Duration - 21:29
Processed 44.2 GB
Read 1.7 GB
Transferred 782.8 MB

EVENT- SERVER SNAPSHOT TAKEN - KEPT 3 DAYS THEN DELETED

Post Snapshot Deletion
Backup Type - Reverse Incremental - same server
Duration 2:07:00
Processed 74.5 GB
Read 72.5 GB
Transferred 1.1GB

This behavior occurs across my entire server infrastructure. Generally during patching I snap the servers, patch them, monitor behavior, then delete the snapshot.

Every month after I delete the snaps-shots my backup jobs take all night and all of the next day. From what I can see they are rereading all the data again.

I did contact support and they said this was due to change block tracking. I contact VMWare and they just told me more about change block tracking.

At this point I just want to know if I am anomaly, following bad practice or have something configured wrong. Like to know if any else has seen this behavior.

thanks
J>

[Vitaliy S.]

It seems like CBT data gets reset and that is why Veeam is forced to re-read VM data again. Did VMware team explain you why this happens? Do you see the same behavior of all VMs?

[veremin]

Additionally, the time interval between full and incremental run is about 3 days. So, I’m wondering what of type of it’s: Is this a particular VM with high change rate (SQL, Exchange, etc)? Are there any antivirus and/or defrag activities being performed at this server?

If so, it’s expected that almost all of the blocks have been changed within 3 days and VB&R needs to reread them again.

Hope this helps.
Thanks.

[jholton@nmfa.net]

Thanks for the replies.

Vitaliy - yes that is the exact behavior. It appears that the CBT is getting reset and Veeam re-reads everything again. VMWare pointed to Veeam, Veeam support pointed to VMWare. I was hoping to get more information about CBT from VMWare but they just kept sending me articles about how to turn it on. I was hoping someone on the forum has seen this behavior in their own environment. Otherwise I will keep thinking something is not configure correctly on my end.

v. Eremin - There is no change in the type of job being run. Every night I run a reversed incremental backup. The times remain normal until I wait X Days until deleting a snapshot. Then it seems a re-read is triggered. This effect is equal for all servers. Whether it is an Exchange server or a file server. Actually the file servers are the worst because they have the largest disks. We do run antivirus on all servers.

Thanks to both of you thinking on it more I may run some tests to see what X is. Perhaps there is a tipping point for the re-read trigger to occur. I will post any findings.

J.

[veremin]

Then, for the purpose of testing It might be worth taking snapshot manually via vSphere client and seeing whether it resets the CBT or not. If yes, the issue should be definitely addressed by VMware.

Hope this helps.
Thanks.