Exchange 2010 DAG (2 nodes) where to put Journaling DB's?
Hello,
There seems to be a lot of confusion about DAG and Symantec EV. I think the whitepaper "Technical Whitepaper - Enterprise Vault 9 Archiving from Exchange 2010.pdf" from Dan Strydom does solve some questions but not all. As a point the "Jounraldatabase1" on page 8 is hosted on MailboxServer1 has a copy on the DR MAilboxserver3 but there is not copy on Mailboxserver2.
Lets start with a sample that most middle site customer have and is also mentioned in the paper above on page 8.
2 x CAS
2 x Hardware Load Balancers
2 x MBX Server with DAG
MDB1 (Active on Server 1)
MDB2 (Active on Server 2)
Lets also asume they have a Backup Solution which handles the DAG and sees the DAG (Netvault and Netapp SME).
1) Microsoft says a Database which resides on a DAG Server should be DAG activated. (That would not be in sample with JOURNALDATABASE1). Either its DAG or non DAG but nothin between. But then you would have the IOPS + the Log traffic between the DAG additional.
2) Microsoft says that Journaling will give 15-20% more IOPS with Journaling
and to keep it in seperate DB
Some things we still don't understand or how we understand it before we asked here:
a) Best would be the Journaling DB to be in the regular DAG database. However you would have the 15-20% IOPS increase.?
b) Better would be per Server a seperate standalone DB for the Journaling. This will not failover. This will give EVENTS and error because the DB is NOT DAG activated and resides on a DAG Server. > Backup products may hang at a selection point for backup or restore because the can't handle the special DB (Let's just asume)
c) If we make a seperate standadlone Journaling Server to get rid of 1) IOPS 2) Problems with Events 3) Asumed problems with Backup components THEN we have the Problem that single Journaling Server is NOT Redundant.
How do you solve this in practice and whats the best way Symantec would recommend?
If you look at the Forums many customers seems to have problems with that. We just want to understand why and best practice.
Thank you for any help and your time. Any comment welcome.
Regards
Mike
http://www.symantec.com/docs/HOWTO57591
http://www.symantec.com/docs/TECH144726
Hello Mike.
Here is how we do it. This is in a relatively large environment.
We have our mailboxes distributed over dag's, with 1 replica only. (ie 1 prod, 1 dr).
We have a seperate database (not dag) that hosts the Journal Mailbox on the least busy Exchange server. In addition to that, we also have a non-dag database on another Exchange server, hosting a 'backup' journal mailbox, and a 'backup journal' mailbox on one of our DR servers. We archive from that live mailbox. I do not see the benefit of having a journalmailbox in a dag, because when using archiving, it is 'messages in, messages out'. There will be lots of log-files, without use, because items are removed from the mailbox almost immediately due to the archiving taking place.
We use premium journaling (i.e use rules to define who is journaled and who is not). If something happens with the server that hosts the journal mailbox, we have a powershell thingie that changes the journalmailbox being used. The 3 journalmailboxes are having seperate tasks each.
We monitor the journal mailbox by having a powershell query send an 'item-count result' every 2 to 3 hours. If the count goes up, investigation is done. If necessary, journaling is reconfigured to point to one of the backup mailboxes, and the belonging task is started. My environment does not generate many items in the journal mailbox, so I can do for 7 to 10 days without journal mailbox archiving.
If you're journaling generates a massive amount of mail daily, think of 500.000+, I would definitely have a seperate 'journaling' exchange server. I would also make sure I have a backup one, with backup journal mailboxes. I would then also define a few tasks (instead of having 1 task do all jnl mailboxes), but that is up to you and your environment.
