Tip on how to speed up snapshot commit for Exchange DAG

[swbcwes]

Like many here, I was struggling with issues with my Exchange 2013 DAG backups, mainly due to storage contention during the snapshot commit (remove snapshot). We have a passive node that we are using strictly for backup. I first tried to put the node into maintenance mode, which prevents the log files from being played into the database. This helped a little, by keeping the database the same, but I still had to deal with all the log files that were copied during the backup. I decided to try suspending the database copies during the backup and that made a huge difference.

Time to delete snapshot on Passive Node:
Normal: 1.5 to 3 hours
Maintenance Mode: 45 minutes to 1 hour
Mailbox Database Copy Suspended: 2-3 minutes!!!

Note: I tested a restore of an Incremental Backup using this method and was able to restore and view and restore emails with the Veeam Explorer for Microsoft Exchange with the database suspended.

I first tried this manually and after it worked, I create a script to do the suspend and resume. My scripts are running on the Exchange Server itself, but you could certainly do this within the Veeam job under the Advanced Settings, Advanced Tab, then Job Scripts.

StartBackup.ps1

Code: Select all

get-mailboxdatabasecopystatus -server passivenode|Suspend-MailboxDatabaseCopy -SuspendComment "Backup" -Confirm:$False

StopBackup.ps1

Code: Select all

get-mailboxdatabasecopystatus -server passivenode|Resume-MailboxDatabaseCopy -Confirm:$False

Disclaimer: Attempt at your own risk. While I have tested this and it is working well for me, attempt first in a lab...yeah...like everyone has one of those. At the very minimum, run through your own backup/restore test to make sure it doesn't impact your ability to do the restores you want.

[Gostev]

Thanks for sharing this with the community, Wes. I am also moving this to the general forum for visibility - and because this should help with snapshot commit speed regardless of hypervisor in use.

[ferrus]

I've been using these scripts successfully for some time, and they've made backing up our passive Exchange 2010 node - incredibly easy.

During the final migration away from our old backup system, we finally enabled the log truncation option in the Veeam Backup job.
Unfortunately, only the public folder logs were truncated - the rest were ignored.

After a lot of testing, we removed the pre/post scripts above - and log truncation occurred across all mailboxes immediately.
That's fine for our current requirements, but is anyone aware of another method of suspending the DB copies - which still permits log truncation?

[foggy]

I'm not sure suspending the node allows for logs truncation, since in this case nodes are not in sync (logs are actually truncated on active node and then replicated back to passive, AFAIR).