1
|
|
2
|
- Establish monitoring benchmarks
- Monitor server services, logged-on users, and server functions
- Use Task Manager to monitor processes and performance data
- Use the System Monitor to monitor page file, memory, processor, disk and
other critical server performance functions and to tune these functions
as needed
|
3
|
- Set up performance logs and alerts for monitoring
- Identify key system elements to monitor for problems
|
4
|
- Monitoring helps you become familiar with normal server performance to
help identify areas of growth, prevent problems, and troubleshoot
problems later
- Plan to monitor a range of activities to better understand a server
|
5
|
|
6
|
- Plan to establish server benchmarks as an effective way to determine
normal performance and distinguish it from problem situations
- Benchmarks can be established by:
- Generating statistics about CPU, disk, memory, and I/O with no users on
the system
- Using performance monitoring to establish slow, average, and peak use
periods
|
7
|
- Gathering performance information for slow, average, and peak use when
new software is added
- Gathering information about growth in server use, number of users,
software use and average time users are on the server
|
8
|
- Monitor server services to determine if any are stopped, improperly
configured, or possibly hung
- Use the Computer Management tool or the Services tool to monitor
services (both can be accessed from the Administrative Tools menu)
|
9
|
|
10
|
|
11
|
|
12
|
|
13
|
|
14
|
- When you monitor services, the default information that is displayed
includes:
- The name of each service
- A description
- Status of the service (started, paused, or halted/not started
- Status type (how the service was started)
- Logged on as (where the service is running)
|
15
|
|
16
|
- Stop a service carefully because other services may be dependent on it
- Check on dependencies by:
- Double-clicking the service in the Services tool
- Clicking the Dependencies tab
|
17
|
|
18
|
- Configure a service’s properties by double-clicking the service in the
Services tool
|
19
|
|
20
|
- To monitor logged on users:
- Right-click My Computer and click Manage
- Double-click System Tools (if necessary)
- Double-click Shared Folders
- Click Sessions
- To monitor resource use, double-click Shares instead of Sessions as the
last step
|
21
|
|
22
|
- Use the Task Manager to monitor:
- Applications
- Processes
- Basic performance
|
23
|
|
24
|
- To start Task Manager:
- Press Ctrl+Alt+Del and then press Task Manager to start the Task
Manager
- An alternative way to start Task Manager is to right-click an open area
of the taskbar and click Task Manager
|
25
|
- When you press Ctrl+Alt+Del after Windows 2000 Server is booted you have
the following options:
- Lock Computer
- Change Password
- Log Off
- Task Manager
- Shut Down
- Cancel
|
26
|
- When you monitor CPU performance, 100 percent CPU utilization is not a
concern if it is temporary, but if it frequently lasts for several
minutes instead of several seconds you should troubleshoot the cause
|
27
|
|
28
|
- If you believe a particular program is causing a CPU bottleneck and want
to monitor CPU use for its associated process:
- Right-click the program in the Applications tab
- Click Go To Process
- Examine the CPU and CPU Time columns for that process on the Processes
tab
|
29
|
- Use the Task Manager to:
- Stop a task (application) or process
- Stop the process tree (the process and the subprocesses it has started)
- Change the priority of a process
|
30
|
- A process’s priority can be set to:
- Low (-2)
- BelowNormal (-1)
- Normal (0)
- AboveNormal (+1)
- High (+2)
- Realtime (+15)
|
31
|
|
32
|
- Use the Realtime priority with great caution because it may cause a
process to completely dominate a server’s resources
|
33
|
- The Task Manager Performance tab enables you to monitor:
- CPU performance
- Memory performance
|
34
|
|
35
|
|
36
|
|
37
|
|
38
|
- Use the System Monitor for in-depth monitoring of all kinds of elements
on a server, called objects
|
39
|
|
40
|
|
41
|
|
42
|
|
43
|
- Specific qualities of objects are measured by counters and instances:
- Counter: Used by the System Monitor, this is a measurement technique
for an object, such as measuring the processor performance by
percentage in use
- Instance: Used by the System Monitor, when there are two or more types
of elements to monitor, such as two or more threads or disk drives
|
44
|
|
45
|
|
46
|
|
47
|
- System Monitor offers three tracking modes:
|
48
|
|
49
|
- You can run System Monitor inside Microsoft Word to easily capture and
print specific charts and information
|
50
|
|
51
|
- Use the System Monitor Memory and Paging File objects to monitor memory
and page file performance
|
52
|
|
53
|
|
54
|
|
55
|
|
56
|
|
57
|
- A page fault can occur in a kernel process when a page location in
virtual memory is lost or corrupted. If this happens, Windows 2000 may
crash with a Stop message. Try rebooting to determine if the problem
recurs. If it does not, then the problem was simply transient. If it
does, employ memory diagnostics to look for a defective memory module.
|
58
|
- Monitor the effect of software on memory use, such as looking for
leaking memory:
- Leaking memory: Failing to return memory for general use after a
process is finished using a specific memory block
|
59
|
|
60
|
|
61
|
- Look at three important components when you monitor processor load:
- Percent of time the processor is in use
- Length of the queue continuing processes waiting to run
- Frequency of hardware interrupt requests
|
62
|
- Plan to gather benchmarks on the frequency of hardware interrupts to
have comparative data for troubleshooting hardware problems later on
|
63
|
|
64
|
|
65
|
|
66
|
- The Disk Performance Statistics Driver (Diskperf) must be set up to
enable disk monitoring
- Run Diskperf from the Start button Run option (or from the Command
Prompt window) and specify the desired switch, such as Diskperf -y which
installs the driver and System Monitor counters for physical and logical
drives
|
67
|
|
68
|
|
69
|
|
70
|
|
71
|
|
72
|
|
73
|
- Use the Terminal Services object in System Monitor to monitor:
- Active sessions
- Inactive sessions
- Total sessions
- Use the Terminal Service Session object to monitor selected sessions or
all sessions for their affect on the server load
|
74
|
|
75
|
- Monitor file system caching so that you can add memory or adjust the
memory allocation as needed
- File system cache performance is monitored in terms of cache hits and
misses:
- Cache hit: A situation in which the data needed by an application is
found in cache
- Cache miss: Occurs when the desired data is not found in cache
|
76
|
|
77
|
- Use the Network and Dial-up Connection tool to tune system cache
|
78
|
|
79
|
- Performance log: Tracks system and network performance information in a
log that can be viewed later or imported into a spreadsheet, such as
Microsoft Excel
- Alert: Provides a warning of a specific Windows 2000 Server system or
network event — the warning is sent to designated users
|
80
|
- There are two types of performance logs:
- Counter log: Traces information on specific System Monitor objects in
selected intervals
- Trace log: Traces specified events when they occur
|
81
|
|
82
|
|
83
|
- Counter logs can use large amounts of disk space and slow performance,
thus set the snapshot intervals relatively high, such as at 15 seconds
or higher when you monitor for 4 hours or less
|
84
|
|
85
|
- Use an alert to provide a warning each time a particular event occurs,
such as when the CPU is at 100 %
|
86
|
- Run the Task Manger, System Monitor, performance logging, and alert
monitoring sparingly because these all can place an extra load on a
server
|
87
|
- Create a set of benchmarks so that you have comparative information that
helps you prevent problems and solve problems after they occur
- Monitor server services to make sure that all services are properly
configured and started
- Use the Task Manager to monitor applications, processes, and performance
|
88
|
- The Task Manager also enables you to tune a server by tuning process
priority
- System Monitor is a versatile tool that can be used to monitor all kinds
of objects, counters, and instances
- Performance logs use System Monitor capabilities to track information
for later reference while alerts give you instant notification about a server condition
|