Microsoft 1 Windows 2000 Objects and Counters to Monitor

Chapter 4: Enterprise Monitoring 59

If you are running Instant Messaging in your organization, you may find that the

organization quickly becomes as reliant on Instant Messaging as it is on e-mail. It is

therefore important that you monitor Instant Messaging Counters. You should

examine the following:

◆Current Online Users – This shows the number of users logged on to the server. Exam-

ining this parameter over time helps to determine the actual take up of Instant Messag-

ing in your organization and therefore help you to scale it properly across multiple

servers.

◆Current Subscriptions – This shows the number of subscription notifications sent to the

server by the Instant Messaging client. A subscription notification occurs when a user is

added to the contact list. This gives an indication of how heavily clients are using

Instant Messaging.

◆Inbound Subscribes/sec – This shows the average number of subscribes/second. If this

figure is low but the usage of Instant Messaging is high, it could indicate an over-

worked Instant Messaging server.

The Recipient Update Service (RUS) plays a crucial role in the day-to-day operations of

Exchange 2000 because it is responsible for keeping e-mail addresses and membership of

address lists up to date. You should measure the Address List Queue Length when examin-

ing the RUS:

The Address List Queue Length shows the load the Recipient Update Service is under. If

this value is consistently high compared to your baseline, you should seriously consider

upgrading the server that has this role, or transferring the role from a weak or overloaded

server to a more powerful one.

Windows 2000 Objects and Counters to Monitor

A heavily used Exchange 2000 server may have a number of bottlenecks. Simply monitor-

ing Exchange 2000 Server performance objects and counters in isolation will not give you

information about the condition of the server itself. You will need to monitor for bottle-

necks in the Disk Subsystem, Memory, Processor, and the Network Subsystem. For ex-

ample, in many cases there will be multiple instances of disks and processors, so make sure

that you monitor all instances (that is, each disk or each processor). Table 4.2 shows which

objects and counters it would be most useful to monitor, along with any specific notes

regarding Exchange.

Note: When monitoring disk counters, you need to enable them to start at boot, using the

diskperf –y command.