Microsoft recently released Exchange Server 2010 RC1 to the public. While some minor things may change prior to the final release of the product, we can assume that most things will stay the same. The Exchange team has made a large investment into archiving, compliance, discovery, and data retention in this release, adding features such as:
- Archive Mailbox
- Granular Retention Policies
- Multi-mailbox search
- Enhanced transport rules processing
Several of the most important new features that are tied to archiving do not appear to be related, at first glance. These include changes made to the Exchange database itself. The database is called ESE (pronounced “easy”), which stands for Extensible Storage Engine. Due to changes in layout and architecture, the database engine has experienced a 70% reduction in input-output (I/O) operations when compared to a similar workload versus Exchange 2007. When compared to Exchange 2003, this reduction in I/O is over 90%!
The goal was to allow lower-performing SATA disk (that is, inexpensive workstation class disk drives) to be used for Exchange database storage. Microsoft was successful in this goal. This means that with Exchange Server 2010 it is now possible to have many times the capacity in cheap disk, as compared to the much more expensive SCSI and SAN disk that was previously required for good performance.
This will undoubtedly cause an explosion in mailbox sizes and in the amount of data to be backed up. Beware! Cheap disk will fail more often (in general) than expensive disk. Your backups and your email archives will be even more important now than they were in the past.
One of the changes to ESE that allowed for this reduction in I/O was to remove any dependence of one mailbox upon any other mailbox. In prior versions of ESE, that capability supported something known as SIS – Single Instance Storage. In brief, if a user sent an attachment to two other users that shared the same mailbox database, only one copy of that attachment was actually stored in the database – the recipient users had pointers to that attachment, which was placed into a shared storage area. In Exchange Server 2010, SIS is gone. Every user that receives an attachment has their own personal private copy of it in their mailbox. While Microsoft did add an offsetting feature (attachment compression), it may not be as effective as SIS was in some environments. In order to regain SIS in Exchange Server 2010, a company will be forced to use third party products.
The addition of Archive Mailboxes is also an interesting feature for Exchange Server 2010, albeit also a limited one. Given that it’s now considered OK to put Exchange databases on workstation-class disk, and that Standard Edition does not have any limits on mailbox database sizes, the need for PSTs has become less of a requirement. If you associate an archive mailbox with a primary mailbox, then you can load all of the PST data into the archive mailbox. The archive mailbox is backed up and restored along with the primary Exchange mailbox.
However, in a “what were they thinking?” – the archive mailbox is required to be in the same mailbox database as the primary mailbox. Therefore, you cannot segregate the storage requirements for archive mailboxes to be separate from the storage requirements for primary mailboxes. Also, you can only view the archive mailbox from within Outlook 2010. Down-level clients need not apply. Again, for storage segregation, a company will be forced to use third party products.
Finally for today, multi-mailbox search has been added and is fairly quick and the search criteria are pretty comprehensive. However, the output of the search goes to a destination mailbox and the search is always in batch mode. For real-time search, a company still must use third party products.
In summary, Microsoft has added a great deal of capability in the archiving, compliance, discovery, and data retention features available in Exchange Server 2010. However, for ease-of-use and for certain basic capabilities, a customer will still need to evaluate what is available in third party solutions.