Email archiving and data retention have become increasingly important to businesses of all sizes. Although Microsoft acquired hosted archiving service FrontBridge in 2005 (a service they still offer), the Personal Archive feature in Exchange Server 2010 is Microsoft's first foray into on-premises message archiving. The feature set for Personal Archive covers basic archiving; although other products can offer more functionality or flexibility, one of the key Personal Archive selling points is its deep integration with both Exchange and Microsoft Outlook 2010. (Integration with Outlook 2007 is more limited, and there's no integration with Outlook 2011 for Mac).

In Exchange 2010 RTM, archive mailboxes had to be in the same mailbox database as the primary mailbox. In other words, if you had 1,000 mailboxes in a mailbox database, and each had an archive mailbox, your database would have 2,000 mailboxes. This approach was easy to understand and easy for Microsoft to implement, but it wasn't very flexible.

As is often the case, this lack of flexibility turned out to be a deployment blocker for many customers. In addition, the need to collocate primary and archive mailboxes made it more difficult to design mailbox servers for scale. One of the key advantages of the Personal Archive feature was the promise that you'd be able to give users a much larger amount of archive space, but that was hard to do when archive and primary mailboxes had to compete for resources on the same mailbox servers.

In Exchange 2010 SP1, Microsoft removed this restriction. Personal Archive mailboxes can now be on any mailbox database in the organization. They don't have to be on the same server, or in the same database availability group (DAG), or on the same network as the primary mailbox. In fact, they don't even have to be on your servers; it's now supported to put Personal Archive mailboxes on Exchange Online while the primary mailboxes remain on your servers -- or vice versa.

Given that you can now put archive mailboxes anywhere, where should you put them? The answer: It doesn't matter. From a sizing and scaling perspective, you should treat archive mailboxes just as you would treat regular mailboxes.

For example, let's revisit the 1,000-user scenario I mentioned earlier. If you want to give users 5GB mailbox quotas for their primary mailbox and 25GB archive mailbox quotas, you have 5TB of primary mailbox data and up to 25TB of archive data -- but you won't start with all 25TB of archive data at once (unless your users have unusually large existing archives.) If you use Microsoft's Exchange 2010 Mailbox Server Role Requirements Calculator, you can treat this set of mailboxes exactly as you would 2,000 mailboxes: 1,000 with a 5GB quota and 1,000 25GB mailboxes.

The same rule applies if you're putting archive mailboxes on dedicated servers: Use Microsoft's sizing recommendations in the calculator as though you were sizing any other mailboxes. You can find out more about the calculator in the Exchange 2010 Mailbox Server Role Requirements Calculator blog post, and you can download the calculator from TechNet.

Following this approach is a bit of overkill because users won't spend as much time with the archive mailboxes and thus are unlikely to generate as much load as they will from primary mailbox access. However, it's better to have too much capacity than too little; it's a safe bet that your users' mailboxes will grow into any capacity you might over provision.