What's the solution?
Several alternative approaches come to mind. However, each has its drawbacks. Migrating all mail in advance of the migration date is hardly the answer. After all, mail migration takes all data from the old mailbox and moves it to the new mailbox. Advanced mail migration will miss all the newest sent and received mail. Furthermore, all the recent activity in the legacy system, such as movement of mail items between folders, and deletion of mail items will be lost. Migrating only a small subset of mail isn't the answer either; most users, and in fact organizations, need all mail to be preserved.

The solution to this dilemma is to perform the migration in increments. The bulk of the data can be processed prior to a final phase, which migrates only the most recent data, and therefore preserves all of the users' most recent changes.

What's incremental mail migration?
The incremental mail migration concept is extremely simple: mail is migrated in two or more stages. Each migration stage is responsible for mail within a certain range of dates. Every stage but the final stage migrates mail beyond a certain age, so that during the final migration stage only a small portion consisting of the most recent mail is being migrated. Thus, the final stage is completed within the allotted amount of time.

While the concept is simple, implementation can be challenging. First and foremost, the mail migration software must support date-range or age filtering. Second, it must be easy to set up, initiate, and track, various stages of the migration. Finally, the setup must be sufficiently flexible to support the specific requirements of each individual migration project.

Using absolute Date-Range filters
Date-range filtering is the key. Our particular tool supports both relative (by age) and absolute (by start/end date) date-range filtering. Relative migration can be applied to incremental migration; however, its overall functionality is so broad that it exceeds the scope of this article. Instead, I'll concentrate on absolute date-range filtered migration, i.e. migration of data items earlier than a specified starting date, and older than a specified ending date.

Using absolute date ranges, specific date ranges can be assigned to individual users, groups of users explicitly, or to all users implicitly. These date ranges may also be applied for various data types. CMT, for example, supports different date ranges for Mail, Calendar, and Task/ToDo data.

In other words, not only data of a specific range can be migrated for a group of users, but also data of a specific type within a specific range, to allow the administrator to carefully customize each increment of the migration. For example, an incremental migration could be set to migrate Mail data between January 1, 1901 and December 31, 2004, Calendar data between January 1, 1901 and December 31, 2002, and Task data between January 1, 1901 and June 30, 2003.

Furthermore, customizable scripts are a good way to facilitate the assignment of date ranges for migration stages on any number of users. With such scripts, there's no need to remember which ending dates were assigned to each group of users; the scripts use the end dates of the previous migration stage as a basis for start dates for the next migration stage. This way, users who are grouped together can be easily processed together through any number of incremental migration stages.