Managing the Store & DAG: EXCERPT from Microsoft® Exchange Server 2013 Inside Out

Workers, controller, and memory

As part of the rewrite to managed code, Information Store processing is now separated into the work done by a Store controller process and a set of worker processes, one for each database (active or passive copies) mounted on a server. Each worker process deals with the transactions flowing into its database and does the work required to keep the database healthy, such as background maintenance. The major advantage gained by moving work into a set of worker processes is that any problem that occurs is automatically isolated to a single process and therefore can affect only the database to which the process is connected. Of course, if the controller process dies for some reason or is terminated by an administrator ...