Sams Teach Yourself MCSE Windows NT Server 4 in 14 Days
(Publisher: Macmillan Computer Publishing)
Author(s): David Schaer, et al
ISBN: 0672311283
Publication Date: 12/15/97

Chapter 10
Optimizing Your Server

by Theresa Hadden and David Schaer

10.1. Overview

An MCSE should understand how to properly monitor systems for two reasons. The MCSE must perform proactive system maintenance by identifying potential system bottlenecks. Also, when faced with a problem the MCSE must be able to identify the problem’s symptoms, evaluate their cause, and then develop a solution that removes that cause. This chapter will address these issues and present appropriate solutions.

10.1.1. Objectives

This chapter instructs you in how to use the Performance Monitor to evaluate the performance of system objects. Additionally, the chapter shows you how to identify the most common physical objects and then evaluate them under a given set of circumstances. This chapter shows how to

•  Use the Performance Monitor to assess the performance of various functions.

•  Use Network Monitor to watch network traffic.

•  Identify performance bottlenecks.

•  Choose the appropriate course of action to resolve performance issues.

Remember that your objective will be to identify a bottleneck, which is the physical object responsible for degraded performance, and then evaluate how to decrease its negative effect on performance.

10.1.2. Fast Facts

The following list of facts is a concise picture of the information presented in this chapter. It acts both as an overview for the chapter and as a study aid to help you do any last-minute cramming.

•  The Performance Monitor is the primary tool for monitoring system performance.

•  Use thresholds, which you set in the Alert view, to generate a response message or log an error to the event log.

•  You commonly would use the Log view to establish a baseline.

•  Processor utilization is high if it constantly exceeds 80 percent.

•  Track multiple processors collectively by selecting the System object and the counter %Total Processor Time.

•  The addition of memory is the generic fix-all solution in NT; high pages/second can indicate the need for more RAM.

•  To track Physical Disk utilization, you first must enter DISKPERF –Y (-YE if you are tracking a stripe set) and then reboot the system.

•  Multithreaded applications benefit most from multiple processors.

•  Track the traffic on the physical segment in the Performance Monitor by selecting the Network Segment object and the %Network Utilization counter.

•  Track individual process statistics in the Performance Monitor by selecting the Process object.

•  Collectively monitor services that are not listed individually.

•  Use the object Network Interface to track the statistics of an individual network interface card.

•  The Process Viewer from the NT 4.0 Resource Kit provides you with full details of each active process.

•  Use the Network Monitor tool, which is a protocol, or packet, analyzer, to capture and analyze network traffic.

•  Use the Network Monitor Application component to display and save captured data.

•  You must install the Network Monitor Agent, which allows for the capture of data on a remote subnet, on the computer that will capture data.

•  There are four sections to Network Monitor’s display: Graph Pane, Session Statistics, Total Statistics, and Station Statistics.

•  You can apply filters in Network Monitor either to the data as it is captured or to limit the amount of data displayed from a captured file.

•  There are two ways to display the Task Manager. Either press Crtl+Alt+Del and then select Task Manager, or simply click the taskbar with the right mouse button.

•  The Task Manager can report the Process ID (PID), CPU utilization, and memory utilization for each individual process.

•  The Process Viewer can retrieve detailed information on the memory allocation of any active process.

10.2. The Performance Monitor

The NT Performance Monitor is the administrator’s primary tool for evaluating the performance of a system running NT Workstation or Server. The Performance Monitor primarily is for the measurement of a given system’s performance, but it also provides some rudimentary information on network performance.

The Performance Monitor has four major views that you can use to analyze system objects: Chart, Alert, Log, and Report. Each view simply presents data in a different way. This data can be either from the local computer or from remote computers. In addition to selecting which objects and counters to monitor, you can set how frequently you update your counters.

After you have configured the Performance Monitor’s attributes, save the settings to use again later.

10.2.1. The Chart View

When monitoring current activity, you most commonly would use the Chart view, as shown in Figure 10.1, to get a quick overview of system performance in real time. Two types of charts are available: graph and histogram. This view is most helpful when you need a quick look to determine where a problem might be.

When using the graph mode, you can customize line color, width, and style to optimize viewing and printing. You also can edit histogram bars.

Figure 10.1.  You can use the Chart view in the Performance Monitor to show %Processor Time.

Select a legend and press the backspace key to highlight that line.

10.2.2. The Log View

The Log view, as shown in Figure 10.2, shows the log file’s name and status. A system administrator can use the Log view to select a series of objects and track them over an extended period of time. The administrator can change the time interval without closing the log file. After the log has captured the requisite amount of data, the administrator then can view the data in chart, report, or alert format by changing the view and setting the Data From option.