Exchange 2010 DAG DBs Seeding Option for DRP Implementation

Hi there,

• Exchange 2010 SP3
• 1 stretched DAG (between main datacenter and Disaster Recovery Site)
• DAG with 3 member servers: 2 members at our main data center site. We just added one DAG member to our Disaster Recovery (DR) site in order to achieve site resilience
• 2 HT/CAS Servers in a NLB Array at main datacenter
• 1 HT/CAS Servers in Disaster Recovery Site

GOAL: Activate site resilience for our Exch 2010 Email System

WHAT WE DID:

We have large databases at our main datacenter (over 230GB). Since usually seeding DBs over 200GB across a WAN increase the chances of seeding failure, we opted for the following approach:

1. Created a brand new DB for each of the existing DBs we had
2. Then, we created passive copies of each of the new empty DBs on the two mailbox servers in main datacenter and also on the mailbox server in DR site. Doing it this way we did not have to worry about seeding huge amounts of data across the WAN
3. Then the new DBs were replicated across the WAN almost instantly as the new DBs were obviously empty
4. The next step is to start moving mailboxes from old DBs to new DBs. Once again, doing it this way, data will be replicated from main datacenter to DR site progressively as mailboxes are being moved rather than trying to seed huge DBs over the WAN
5. Once we finish moving mailboxes from old to new DBs, we will delete the old DBs to free up space on our storage BOX

QUESTION:

I need some feedback from you experts

As you know, moving mailboxes generate a bunch of transaction log files. Based on the number of mailboxes I need to move from old to new databases, I was considering enabling (temporarily and only during the mailboxes move) circular logging on the new DBs as otherwise  I will have to perform backups constantly in order to flush the transaction logs (my transaction logs partitions are small in size). I did that when I moved mailboxes from Exch 2007 to Exch 2010 and I did not have any issues. The thing now is that a DAG is involved and I’m not quite sure how the use of circular logging will affect the replication and the passive copy of the DB in the DR site taking into consideration that I have setup lagged copies to DR site with a delay of 24 hours.

According to the theory, “Circular Logging allows Exchange to overwrite transaction log files after the data that the log files contain has been committed to the database” so I think I should be ok but wanted to get a feedback from someone that had probably do over this particular scenario

As stated above I'm planning to turn on circular logging on the new DBs...is this the way to go or do we also need to turn on circular logging on the old DBs?

Thanks in advance for your feedback!

 FT

FT