High Performance NT4 Optimization and Tuning:Introducing The Performance Monitor

Table of Contents

Not all performance object counters are useful for specifying alert conditions. The object counters I find most useful, and which you may find useful as well, are summarized in Table 4.1.

**Table 4.1** Performance object counters for alerts.
Object	Counter	Instance	Condition	Description

LogicalDisk	% Free Space	Per Drive	Under	Useful for notification in a low free space situation. Specify the minimum percentage of free space per drive in the Alert If value field. Set this value to 10 or higher for useful warnings. Any value below 10 may notify you too late for your users’ comfort.
LogicalDisk	Free Megabytes	Per Drive	Under	Useful for notification in a low disk space situation. Specify the minimum free space in MB per drive in the Alert If value field. On a large drive (over 1GB), set this value to 100 or less, if you want to run on the ragged edge.
Memory	Pages/Sec	Per Paging File	Over	Set this to five for a notification that the system is paging too often and to indicate whether a memory upgrade may improve performance if consistent alerts are sent. Set this value to 10 to be notified that a memory upgrade is required as the system is definitely paging too much.
Network	Current Interface	Per Network Bandwidth	Over Transport Protocol	Set this value to 50,000,000 (in bits per second) for a notification that your network transport protocol has reached 50 percent capacity.
Network Segment	% Network Utilization	Per Network Adapter	Over	Set this value to 50 for a notification that your network segment has reached 50 percent of its carrying capacity.
Redirector	Network Errors/Sec	N/A	Over	If you have an error, you should spend more time isolating it, so this value should be set to one.
Server	Error System	N/A	Over	Any value over one indicates a problem with the server. If detected, you should examine the system log for the cause of the error or isolate the cause by spending some time with the Performance Monitor, as specified in Chapter 6 “Tools and Techniques” in the section titled “Finding Network Bottlenecks.”
Server	Errors Logon	N/A	Over	This item can be used to notify you that someone is attempting to hack into your system. A good threshold value is from 3 through 10.
Server	Errors Access Permissions	N/A	Over	This item can be used to inform you of potential users trying to access files they should not be attempting to open in order to gain access to privileged data. A good threshold value is from 3 through 10.
Server	Pool Nonpaged	N/A Failures	Over	This item should be set to one, as this indicates a lack of nonpagable memory for the server service. An error indicates that the server will be unable to process a client request for lack of resources. This value also indicates a physical memory shortage.
Server	Pool Paged Bytes		Over	This item should be set to one, as this indicates a lack of pagable memory for the server service. An error indicates that the server will be unable to process a client request for lack of resources. This value also indicates a physical memory shortage or that the Pagefile setting is too low.
Server Work	Work Item Queues	Per Network Shortages	Over Adapter	This value should be set to zero, as any indication represents a problem with the server service. If errors are encountered, the MaxWorkItems value should be increased.

Summary

This chapter discussed the basics of using the Performance Monitor to create charts, logs, alerts, and reports, and it gave you an idea of how to use these features. This chapter also included specific performance object counters for you to generate alerts to aid you in keeping your network server up and running.

The goal of this chapter was not to explain every detail of the Performance Monitor. Instead, the goal was to give you a basic understanding of the Performance Monitor’s features and provide you with the ability to use the Performance Monitor in your day-to-day activities. In future chapters, we will refer to the Performance Monitor again when we discuss specific optimizations for various subsystems (such as your processor, memory, I/O, and networking subsystems).

Table of Contents