In Exchange 2010, using address book policies and third party custom send connectors I have shared our Exchange resources with other related but separate organizations.

To date, I've always created separate database and log volumes for each organization, sometimes, multiple databases for large organizations. This has resulted in widely varying database sizes. I generate reports and other management tasks based on a value in CustomAttribute1 that identifies what organization an account belongs to. But, I also run some daily scripts based on the mailbox's database placement.

I'm wondering if maybe I should abandon the database for exclusive organization model and allow the mingling of accounts from all organizations into multiple databases. It would require a reworking of some of my scripts. This might help in keeping the size/user count similar between the various databases.

Is there a best practice or consensus on how to create database and log volumes for related but separate organizations in one Exchange org?

• Edited by Thunderpup Wednesday, August 12, 2015 7:23 PM

A follow up question: We use iSCSI for all of our storage and to date, I've  built a separate volume for every database and a separate volume for every set of logs. We do that because a number of years ago, with multiple databases on one volume, a single database consumed the drive and it took a couple of days to recover. Is this no longer best practice? Should I return to putting more than one database on a volume?

The Exchange 2013 Server Role Requirements Calculator can help you decide how many LUNs you really need.  There is no compelling reason to create individual LUNs per database unless you need that for I/O, and you probably don't.  In my opinion it's better to use as few LUNs as your I/O requirements allow because your free space is available to any database that grows and your performance is pooled among the drives.