One of the most important aspects of
monitoring performance is that it should be done over a given period of
time (referred to as a baseline). So far, we have discussed how you can
use Performance Monitor to view statistics in real time. We have,
however, also alluded to using Performance Monitor to save data for
later analysis. Now let's take a look at how this is done.
When viewing information in Performance Monitor, you have two main options with respect to the data on display:
View Current Activity
When you first open the Performance icon from
the Administrative Tools folder, the default option is to view data
obtained from current system information. This method of viewing
measures and displays various real-time statistics on the system's
performance.
View Log File Data
This option allows you to view information that
was previously saved to a log file. Although the performance objects,
counters, and instances may appear to be the same as those you saw
using the View Current Activity option, the information itself was
actually captured at a previous point in time and stored into a log
file.
Log files for the View Log File Data option are
created in the Performance Logs and Alerts section of the Windows
Server 2008 Performance tool.
Three items allow you to customize how the data is collected in the log files:
Counter logs
Counter logs
record performance statistics based on the various performance objects,
counters, and instances available in Performance Monitor. The values
are updated based on a time interval setting and are saved to a file
for later analysis.
Circular logging
In circular logging, the data that is stored
within a file is overwritten as new data is entered into the log. This
is a useful method of logging if you only want to record information
for a certain time frame (for example, the last four hours). Circular
logging also conserves disk space by ensuring that the performance log
file will not continue to grow over certain limits.
Linear logging
In linear logging, data is never deleted from
the log files, and new information is added to the end of the log file.
The result is a log file that continually grows. The benefit is that
all historical information is retained.
Now that we have an idea of the types of functions
that are supported by the Windows Server 2008 Performance tool, let's
move on to look at how this information can be applied to the task at
hand—monitoring and troubleshooting Active Directory.
In our daily jobs as systems engineers and
administrators, we come across systems that need of our help. . .and
may even be asking for it. Of course you check your Event Viewer,
Performance Monitor, and perform other tasks that help you
troubleshoot. But what is really the most common problem that occurs?
From our experience, we'd say that many times you suffer performance
problems if you have your Windows Server 2008 operating system
installed on a sub-par system. Either the server hardware isn't
enterprise class, or the minimum hardware requirements weren't
addressed. Most production servers suffer from slow response times,
lagging, and so on, because money wasn't spent where it should have
been—on the server's hardware requirements.
Take a look at www.microsoft.com/windowsserver2008/evaluation/overview.mspx
to see the minimum Windows Server 2008 requirements. You have to make
very sure that you follow these minimum requirements.
Would you drive a truck over a glass bridge? No.
Then why would you run an enterprise class server operating system
hosting a mission-critical application such as Active Directory, email,
and messaging on an antiquated desktop system? This seems illogical
when you read it, but in practice, it's common to find budgets squeezed
to the point where your secondary domain controller is running on a
high-end desktop. Just make sure that you consider this when you deploy
a new system. Once you deploy it, open up Performance Monitor and see
if you are having issues just opening and running programs on the
server itself.
It's also common to blame the network first, but it
is usually not the problem at all. Be careful of false positives and
keep your mind focused on finding the root of the problem. If you come
across other problems, document them, but continue to focus on finding
(and fixing) the real issue.
If your enterprise-level servers aren't
running with Redundant Array of Independent Disks (RAID), then you will
most likely need an upgrade on your system hardware. Most
enterprise-level server systems come with RAID as the minimum fault
tolerance you should have on any server of any size. RAID can help you
in a pinch; when you lose a disk—and you will, based on the Mean Time
Between Failure (MTBF)—you can quickly recover with minimal downtime
and no loss of data.
