Page 176

138

Chapter 3

Generic Troubleshooting Tools

The basics behind these systems are that, with some configuration and

guidance, they use different methods of discovering a network. This is done
by starting out with a specified device--a seed device. The seed device polls
all of its interfaces and comes back with the necessary information that
enables the program to move on to the directly connected neighbors. The
polling goes on until an edge is reached and there are no more neighbors to
discover. Both of these systems draw logical topological network maps.

In addition to network discovery, the NMS monitors for device availability

and reachability. If something does affect the connectivity, an alarm is tripped
within the software, and it logs the event and displays an alarm.

Monitoring can also be done on a more detailed level (such as threshold

monitoring). Thresholds can be defined within the software. They tell the
program to trigger an alarm if a specified variable for a given machine
exceeds a maximum or descends below a minimal value. These alarms can be
dealt with in various ways (for example, e-mail or pages may be sent).

The previous examples are just a few examples of what NMS packages

can do. Here is a list of what most management systems try to do:

Availability management This was described previously as network
monitoring.

Network performance management This is done by measuring traffic
loads and other bandwidth-oriented data that can be used to calculate the
network's overall performance.

Network security management This is done by making the NMS the
means by which changes are made to network devices. Because the software
requires the user to log in, it can also track changes made by the user. Security
management can also be done via having a user database within the NMS.
When a network device is accessed or a change is attempted, the user is
authenticated from the user profiles located within the NMS.

Network service simulation In today's networks, it becomes very risky
to test out configuration changes on a live production network. Simulation
software that enables changes to be made off-line and tested before being
implemented is a big part of network management. It gives the administra-
tor the ability to see whether the changes will cause any side effects, without
endangering the applications on the production network.

Policy-based management This has to do with QoS, or Quality of Service.
When an administrator knows that certain applications require more network
resources, he or she can allocate resources accordingly. With policy-based
management, the administrator can see where the most resources are needed
and make it a higher priority that those resources are available when needed.

www.sybex.com