Chapter 10
The deployment of fewer but more powerful Microsoft Exchange 2000 servers instead of more but less potent systems can help your organization to keep the total cost of ownership (TCO) at a minimum. The more users you place on a server, the more you reduce the hardware and software costs per user and administration and maintenance requirements. Storing numerous mailboxes on a single machine also results in decreased message traffic between the servers, which ultimately saves network resources. With Microsoft Windows 2000 Advanced Server or Datacenter Server and Exchange 2000 Server, you can build extremely powerful server machines.
Implementing giant servers requires more than placing several gigabytes (GB) of random access memory (RAM), eight processors, and a terabyte (TB) of disk space into a computer. Among other things, you need to account for normal and peak workload. The latter can quickly consume your available resources, resulting in unacceptable response times and end user complaints. Besides performance, you also must take into consideration unexpected situations, such as power and disk failures. An uninterruptible power supply (UPS) should be standard for any production system, but servers that host thousands of mailboxes also require sophisticated and fault-tolerant storage because availability is a requirement for high-powered servers.
This chapter explains how to design Exchange 2000 servers that require a high level of fault tolerance and system resilience. With an appropriate level of fault tolerance and a clever mailbox store design, you can place several thousands of users on a single machine. Lesson 1 covers the system configuration for standard servers that are not part of a server cluster. Lesson 2 then addresses the design of clustered Exchange 2000 servers, which allow you to increase your systems’ availability.
To complete the lessons in this chapter, you need to