Executive summary
This paper describes a best practice process for consolidating servers and storage in a multiserver
Microsoft® Exchange 2000 Server environment by concurrently migrating to a single server Microsoft
Exchange Server 2003 environment on a storage area network (SAN). The methodology described is
commonly referred to as the “move mailbox” method. It includes installing a new Exchange 2003
server into an existing Exchange 2000 Administrative Group, using the Exchange 2003 move
mailbox wizard to move the user mailboxes from the source Exchange 2000 server to the target
Exchange 2003 server, and decommissioning the old servers.
The HP Customer Focused Testing (CFT) lab performed two Exchange 2000 Server to Exchange
Server 2003 migration tests, which included concurrent server and storage consolidations. The two
test scenarios were based on a typical deployment and map to an appropriate HP Storage array. The
first scenario was based on the HP StorageWorks Modular Smart Array (MSA) 1000, which is an
ideal entry-level SAN-based storage array and was tested with 1,200 users. The second scenario was
based on the HP StorageWorks Enterprise Virtual Array (EVA) 3000, which is an appropriate entry-
level array for a larger deployment and was tested with 3,000 users.
Each of these validated examples should be considered a “proof of concept.” Larger or smaller
numbers of users in the deployment can be used as required, but the steps for migration and
consolidation are the same. Highlighted test results include the following:
MSA1000 testing:
• The migration and consolidation to the MSA1000 was completed without incident, with 1,200
mailboxes totaling 24.57 GB of data moved in 7 hours and 21 minutes, which approximately
equates to a migration rate of 3,343 MB per hour.
• During the mailbox migration to the MSA1000, each end user only experienced an average
interruption of service of 1.48 minutes.
• The 95
th
percentile Load Simulation (LoadSim) score taken during mailbox migrations to the
MSA1000 was 51, which means that during a move operation, the online users would have an
average response time of 51 milliseconds to 95% of Exchange requests. A LoadSim score of less
than 1,000 is considered passing.
• The average LoadSim score taken after each migration was 45, with a load of 1,200 users being
simulated.
EVA3000 testing:
• The migration and consolidation to the EVA3000 was completed with 3,000 mailboxes, totaling
61.44 GB of data, moved in 20 hours and 12 minutes, which approximately equates to a
migration rate of 3,042 MB per hour.
• During the mailbox migration to the EVA3000, each end user only experienced an average
interruption of service of 1.44 minutes.
• The 95
th
percentile LoadSim score taken during mailbox move migrations to the EVA3000 was 59,
which means that during a move operation, the online users would have an average response time
of 59 milliseconds to 95% of Exchange requests. A LoadSim score of less than 1,000 is considered
passing.
• The average LoadSim score taken after each migration was 40.3, with a load of 3,000 users being
simulated.
2
Comentarios a estos manuales