This is the place to start if you are new to the
science of Windows 2003 Performance Monitoring. The purpose of this page is to master the basic
techniques of System Monitor (Perfmon.exe).
We will explore the buttons and discover how to create traces of
server activity.
Mastering the System Monitor controls will help you to
analyse real-time traces as well as loading and interpreting logs. This
page also prepares you for detecting
bottlenecks by capturing key performance counters.
The purpose of this section is to show
you how to launch that vital tool - System Monitor. Microsoft provide two
interfaces, my advice is stick with Perfmon, particularly if your aim is to
collect live data. Perfmon does everything that the MMC version does.
Only use the MMC, Performance Monitor and Alerts when you only need create logs and alerts,
because this version does not allow real-time monitoring.
Perfmon with System Monitor (Best)
The Perfmon executable displays not only the System Monitor but also the Logs,
Counters and Alerts.
Start
--> Programs
-->
Administrative Tools
--> Performance
MMC Snap-in - Performance Monitor and Alerts (Not recommended)
Performance Logs and Alerts (No System Monitor.
MMC
--> Add Remove Snap-in
-->
Performance Logs and Alerts
Normally I recommend the MMC and Snap-ins, but in
this instance I prefer Perfmon because it displays the System Monitor.
From the Console Root you can see the System Monitor, and underneath, the Performance Logs and Alerts. You can launch the Performance
Monitor from the Administrative Tools, or alternatively click
Run, Perfmon (Type), Enter
Once you launch the System Monitor, you get a graph with a trace which reminds me of an
oscilloscope monitoring a heart beat. (See Diagram 1)
Amongst the row of icons is the Display Histogram button, which I recommend
for situations where
you have many objects and you wish to identify the most active.
Finally, there is a 'Report' button. This is one setting I rarely use!
The next icon that I would like to introduce is that big plus in the middle of
the toolbar.
This Add button opens the door to all the counters and objects that you can
monitor.
If you click on the
button,
you will be rewarded by the menus in Diagram 2. The
choice of local computer or network computer is easy; when testing choose local,
but for serious analysis, monitor from another machine so that the very
performance monitoring is not skewing the results.
The selection of Performance object is mind-blowing. Note that the add button always defaults to the Processor.
Your skill lies in selecting
the best instances in the right hand pane. If you select All instances
then the data can swamp you, if your are too selective, then you may miss the crucial counter. Do not worry I will give you plenty of
pointers on choosing the right counters for the right problem. Remember the goal of
this section is just to master the controls.
Two factors will help you to make the best choice of instances, more
experience and a clear mission. So let's get started with the system monitor, I
suggest selecting:
Diagram 2
Guy Recommends: The Orion Network Performance Monitor (NPM) 9.5
Solarwinds' Orion performance monitor is designed for detecting network outages.
This NPM will guide you
through troubleshooting by indicating whether the root cause is a broken link,
faulty equipment or resource overload. Because it produces
network-centric views, it is intuitive to navigate, and as result you can
see easily what's working and what's not.
Perhaps Orion's best feature is the way it suggests solutions. Moreover, if
problems arise out of the blue, then you can configure Orion NPM 9.5 to notify
members of your team what's changed and how to fix it.
One of the best kept secrets of Performance Monitor is the Highlight 'Bulb'
button. The knack of using this technique is to press your keyboard's
cursor up and down arrows, while keeping your eye on the white line on the
chart. (No white line means you forgot to click on the Highlight bulb!)
I once spent 30 minutes tweaking the colors and altering the thickness of the
chart lines. As soon as I discovered this Highlight button, I never looked
back.
Incidentally, if you do need to change the colours, just right click on the
chart under the Color, Scale or Counter and choose properties. I still
have a use for these properties when ever I want to adjust the Scale.
You have a choice of three ways to display the data. Mostly, I stay
with the default Chart view, but sometimes its easier to interpret the data in
the Histogram view.
Let us take Process (not processor) as an example. Our goal is to see
which process is making most calls on the CPU.
Go to the Add Counter, Select Process then click on All instances. You
will get swamped with data but do not worry we will soon control the counters.
If you click on the forth column heading, Instance, you can sort the lines into
alphabetical order. It is no surprise that the _Total is so high, and as
we are not interested in that counter, delete it.
Now switch to the Histogram view (see red ring below), and using the
Highlight button and cursor keys, scroll down to idle time. Delete idle
time too!
Guy Recommends: SolarWinds Engineer's Toolset v10
The Engineer's Toolset v10 provides a
comprehensive console of utilities for troubleshooting computer problems. Guy says
it helps me monitor what's occurring on the network, and the tools
teaches me more about how the system literally operates.
There are so many good gadgets, it's like having free rein of a
sweetshop. Thankfully the utilities are displayed logically: monitoring, discovery, diagnostic, and Cisco tools.
Download your copy of the Engineer's Toolset v 10
Learn
the secrets of which counters to monitor. Master performance monitor
logging,
develop your skills with structured exercises and examples. Print
out a copy to read, while you design logs and alerts to detect network
bottlenecks.
*
Guy
Recommends the Free IP SLA Monitor
The IP SLA Monitor not only helps to discover network bottlenecks, but
also teaches you about router traffic. See how effortlessly this
free monitor analyzes and then displays the IP statistics.
The key to configuring is to select the data most relevant to your
network, for example, ping echo, DNS resolution times, or HTTP
statistics.