Using TDP in a DAG environment

wesd

ADSM.ORG Member
Joined
Jan 6, 2006
Messages
70
Reaction score
0
Points
0
Website
Visit site
My exchange environment will be 2 exchange servers with a total of 8 databases. Each server will host the active copy of 4 databases and the passive copy of 4 databases. I would expect to follow IBM's recommendations and back up the passive copies only but this gives me a problem which I can't get my head around.

Let's say I have a full backup and a couple of incrementals already saved for a database on server A (passive copy) and the database is then made active on server A. The next incremental will now be on server B as this is now where the passive copy is held. If I need a restore a few days later, I don't see how it can work as, on the TSM server, the last full backup and several incrementals will be owned by server A while several other incrementals will be owned by server B.

I would have thought that this is a fairly normal configuration so how are other people handling this?

Ideally, I would like our exchange team to ensure that the active/passive copies are always on the same servers; if a database is made active on the other server I can then start a new Full/Incremental sequence from that point. They, however, will see this as a restriction on the flexibility that DAGs offer and may not agree with the solution and, in any case, it relies on somebody telling me that a database has moved.

I've thought of using the same node name for the TDP client on each server but, on testing, I can see that the tsm server not only stores the data against the TDP node name but includes the baclient node name as part of the filespace name so I'm not sure if this will work?
 
I'm in your exact condition.
It is possible that there isn't a decent step by step guide for the implementation of TDP for exchange 2010 on dag?
 
I logged a call with ibm about this - their answer was that it isn't possible to do what I wanted. Incremental backups must be run on the same server where the last full backup was run. If a database fails over to a different server within the dag, your next backup should be a full backup.
 
I'm in the same situation. Being both the TSM and Exchange admin, I have the luxury of mandating that active copies always sit on the same respective server, but IBM really needs to re-work the TDP agent to just be DAG-aware.
 
I can live with the dag restrictions with one huge exception -

Mail server A runs a full backup and a few incrementals
Database is moved to mail server B which runs a new set of full and incremental backups.
Database moves back to mail server A without your knowledge.

When the database moves back to server A, you need to run a full backup - the incrementals won't be valid since the last full backup was on server B. HOWEVER - an incremental backup will NOT show an error message; it will appear as if it's working until you try to do a restore and find that the server has no record of the incremental!

Backups can be difficult enough to monitor without this.
 
Wesd can i have some information about your configuration?
Didi you configure tdp on both dag nodes?
Do you use powershell script? Or you are using simple cmd like: TDPEXCC BACKUP * FULL /EXCLUDEDAGACTIVE /EXCLUDENONDAGDBS ?
Thanks
 
I have 2 exchange servers in the dag with tdp configured on both. I'm using the tdpexcc backup command to run full/incremental backups and I'm excluding the passive databases (the mail admins have insisted that they want the active copies backed up). It's the simplest possible configuration and yet I need to watch it like a hawk in case a database fails over as I would need to force a full backup in this case.
 
The same for me, the exchange admin want the "excludedagpassive" for the log clearing. The problem is i'm a consulent and i can't always have an eye on his enviroment. We must find a valid solution.
 
Back
Top