Here is something I recently wrote regarding this topic:
The Microsoft Exchange 2000 server, although has the primary function of being a mail delivery system, operates much like a fast database system. It stores information in customized databases, queries that information based on user request, and has the same log and store function as a SQL server. Thus, when we are working on the disk subsystem we should configure and measure the system from almost a database perspective.
The Exchange database system uses a log file and database with a log first -- store second methodology. Information is first written to the log files sequentially in the form of a transaction. Once this is done another process comes along and reads the transaction file. It processes the transaction and stores it in the database. The database is randomly written to and read from.
Then there is the Exchange application itself. This, much as the Windows 2000 OS, is primarily read from the disk. There is no witting taking place in this situation.
Click here to read on.
|
