Performance is a basis for measuring how fast
application and system tasks are completed on a computer and reliability
is a basis for measuring system operation. How reliable a system is
will be based on whether it regularly operates at the level at which it
was designed to perform. Based on their descriptions, it should be easy
to recognize that performance and reliability monitoring are crucial
aspects in the overall availability and health of a Windows Server 2008
R2 infrastructure. To ensure maximum uptime, a well-thought-through
process needs to be put in place to monitor, identify, diagnose, and
analyze system performance. This process should invariably provide a
means for quickly comparing system performances at varying instances in
time, detecting, and potentially preventing a catastrophic incident
before it causes system downtime.
Performance Monitor, which
is a Microsoft Management Console (MMC) snap-in, provides a myriad of
tools for administrators so they can conduct real-time system
monitoring, examine system resources, collect performance data, and
create performance reports from a single console. This tool is literally
a combination of three legacy Windows Server monitoring tools: System
Monitor, Performance Monitor, and Server Performance Advisor. However,
new features and functionalities have been introduced to shake things
up, including Data Collector Sets, resource view, scheduling, diagnostic
reporting, and wizards and templates for creating logs. To launch the
Performance Monitor MMC snap-in tool, select Start, All Programs,
Administrative Tools, Performance Monitor, or type perfmon.msc at a command prompt.
The Performance Monitor MMC snap-in is composed of the following elements:
Overview Screen
Performance Monitor
Data Collector Sets
Report Generation
The upcoming sections further explore these major elements found in the Performance Monitoring tool.
Performance Monitor Overview
The first area of interest
in the Performance Monitor snap-in is the Overview of Performance
Monitor screen, also known as the Performance icon. It is displayed as
the home page in the central details pane when the Performance Monitor
tool is invoked.
The Overview of
Performance Monitor screen presents holistic, real-time graphical
illustrations of a Windows Server 2008 R2 system’s CPU usage, disk
usage, network usage, and memory usage, as displayed in Figure 1.
Additional process-level
details can be viewed to better understand your system’s current
resource usage by reviewing subsections beneath each metric being
displayed. For example, the Memory section includes % Committed Bytes in Use, Available Mbytes, and Cache Faults/sec.
The Overview of Performance
Monitor screen is the first level of defense when there is a need to get
a quick overview of a system’s resources. If quick diagnosis of an
issue cannot be achieved, an administrator should leverage the
additional tools within Performance Monitor. These are covered in the
upcoming sections.
Performance Monitor
Windows Server 2008 R2
comes with two tools for performance monitoring. The first tool is
called Performance Monitor and the second tool is known as Reliability
Monitor. In the previous release of Windows, the Reliability Monitor
tool was included in the Reliability and Performance snap-in. With
Windows Server 2008 R2, the Reliability Monitor tool has been removed
from the Performance Monitor console. The improved Performance Monitor
tool provides performance analysis and information that can be used for
bottleneck, performance, and troubleshooting analysis.
First, defining some terms used
in performance monitoring will help clarify the function of Performance
Monitor and how it ties in to software and system functionality. The
three components noted in Performance Monitor, Data Collector Sets, and
Reports are as follows:
Object—
Components contained in a system are grouped into objects. Objects are
grouped according to system functionality or by association within the
system. Objects can represent logical entities such as memory or a
physical mechanism such as a hard disk drive. The number of objects
available in a system depends on the configuration. For example, if
Microsoft Exchange Server is installed on a server, some objects
pertaining to Exchange would be available.
Counter—
Counters are subsets of objects. Counters typically provide more
detailed information for an object such as queue length or throughput
for an object. The System Monitor can collect data through the counters
and display it in either a graphical format or a text log format.
Instances—
If a server has more than one similar object, each one is considered an
instance. For example, a server with multiple processors has individual
counters for each instance of the processor. Counters with multiple
instances also have an instance for the combined data collected for the
instances.
Performance Monitor
provides an interface that allows for the analysis of system data,
research performance, and bottlenecks. Performance Monitor displays
performance counter output in line graphs, histogram (bar chart), and
report format.
The histogram and line graphs can be used to view multiple counters at the same time, as shown in Figure 2.
However, each data point displays only a single value that is
independent of its object. The report view is better for displaying
multiple values.
Launching
Performance Monitor is accomplished by selecting Performance Monitor
from the Monitoring Tools folder in the Performance Monitor MMC snap-in.
You can also open it from a command line by typing Perfmon.msc.
When a new Performance Monitor session is started, it loads a blank
system monitor graph into the console with % Processor Time as the only
counter defined.
