Exchange DAG Backups

BrimbankCC
BrimbankCC used Ask the Experts™
on
I had this question after viewing how to truncate Exchange 2016 DAG DB transaction log..?.

We are considering implenting a DAG environment.

How does application aware backups using products such as veeam etc impact DAG members that utilise replay lag times and passive dag members?

eg - is there a chance of replay not being healthy if for example the backup job truncates logs but the copy was in a bad state?
Should i be backing up the passive node instead?
If i backup the passive node, what happens to the logs on the active, passive and replay lagged database copies?

All very confusing.
Comment
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
IT Service Manager
Top Expert 2012
Commented:
Everything that you need to know about exchange backups.. Please read the 3 part series article.

https://techcommunity.microsoft.com/t5/Exchange-Team-Blog/Everything-You-Need-to-Know-About-Exchange-Backups-Part-3/ba-p/599928

How does application aware backups using products such as veeam etc impact DAG members that utilise replay lag times and passive dag members?

when a active or passive copy of the dag database backup occurs.. logs from active and passive db are truncated.. lagged copy remains un truncated.

Should i be backing up the passive node instead?

anything copy active or passive is fine.. If you see, people will design DB copy likes 3 mounted in first server and second server will have passive copies.. second server will have 3 mounted and first server will have their copies.. to share the load on the server.

If i backup the passive node, what happens to the logs on the active, passive and replay lagged database copies?

whether you backup the active or passive.. logs will be truncated automatically from active and passive db.. but the lagged copy logs will remain untouched based on you lagged copy settings..
Saif ShaikhServer engineer

Commented:
Any backup software whether it is Veem or any other:

If you back up a lagged database copy, it might take more time to restore the backup because the lagged copy can have more transaction logs to restore and replay. As a best practice, create your backup from a passive synchronous copy and not a lagged copy.

So basically you should taking backup from passive node always and not a lagged copy.

Simple Query and it's Answer:
=========================
For truncation to occur on lagged database copies, the answer must be "Yes" to the following questions:
Is the log file below the checkpoint?
Is the log file older than ReplayLagTime + TruncationLagTime?
Is the log file deleted on the active copy of the database?

I've never had anyone do backups off of the lag copy, but my gut instinct tells me that the other non-lagged copies *may* only end up truncating up to wherever the checkpoint of the lagged copy was.

The other problems I see...

1. If you do a restore then you will be guaranteed you have to replay XX hours/days of log files for a simple restore.

2. Your backup storage is going to require more space due to the excessive # of log files compared to a non-lagged passive copy.

3. If the backup app randomly picks a passive DB then you may end up with really bizare results if it backs up a passive HA copy one day, then a passive non-HA (lagged) copy another day. Your point-in-time backups may actually actually reversed points in time than you'd expect when doing a restore.

Author

Commented:
Thanks all

Do more with

Expert Office
Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.

Start 7-Day Free Trial