"Prevention is better than
cure".
Today, there are many programs
available to assist with network management. These programs can help
identify conditions that may lead to problems, prevent network failures, and
troubleshoot problems when they occur.
Monitoring Applications
One program, "Netcracker", is a
design application that allows network creators to design a simulated
version of their network before putting the real thing together. Another
example is "ConfigMaker", by Cisco. This program allows the designer to
configure network components by using proper operating system syntax and
then tests the implementation as a simulation. These programs are priced in
the thousands rather than the hundreds. However, this is justified by the
amount of time that can be saved by eradicating problems before network
installation. While these programs allow us to build and monitor networks,
they are not a comprehensive solution, and monitoring software should be
used in order to continually check the on-going status of the network. There
are many software monitoring packages available. Sun Microsystems have an
entire range, from small LAN management to Enterprise Network management
packages. "Solstice Site ManagerTM 2.3" is one example: a state-of-the-art
method for managing sites of up to 100 nodes. It simplifies management of
network resources to keep the network running at peak efficiency.
Causes
There are many factors that can
inhibit the performance of a network, leading to a situation called
"bottlenecking". - a sharp and notable reduction in performance. This can be
caused by equipment not capable for the demands that are being placed on the
network. Equipment like network cards, hubs, repeaters etc. Also, the
bandwidth of the cable may not be sufficient for traffic demands. Users can
cause slowdown by playing resource hungry games across the LAN, or engaging
in heavy Internet downloads, like MP3 files and video. The problems can also
be caused by poor LAN organisation, where all nodes populate a single
segment, and therefore a single collision domain. In other words, the
network is like a small room packed with lots of people all talking at the
one time, leading to chaos.
Baselines
It is recommended that a
baseline be established that will assist the network administrator in
monitoring performance. A baseline defines a point of reference against
which to measure network performance and behaviour when problems occur. In
other words, it has to be established what is "normal" for your network,
before it can be determined what is "abnormal". A baseline can be
established by using performance-monitoring software. There may be no need
to buy expensive management software. Users running Windows servers are
provided with integrated management tools at no extra charge. They do not
provide the same range or capability of the higher end solutions, but they
are still powerful tools. These tools allow the administrator to view
various logs that maintain error, security and system information. Other
tools can track processor, disk and memory usage and analyse protocol
performance.
Solutions
Trends gathered by these tools
can indicate the problems previously mentioned, and can help the
administrator prescribe solutions to the problem.
Possible solutions include:
-
Moving to a faster technology by upgrading cable, Interface
cards and components (switches, hubs and bridges).
-
Increasing memory
-
Installing additional CPUs
-
Subnetting (breaking the network into smaller more manageable
chunks using routers).
-
Preventing users from running power hungry games or
applications across the network.
Acceptable Performance
The philosophy of networking is
providing the best service at the cheapest price. It is not difficult to
have a high-performance network. All that is required is the best equipment,
the best technologies, the best methodologies and the best personnel to tie
it all together. However, in the real world this is seldom, if ever the
case, due to costs. Therefore, a trade-off is sought, and ideal performance
gives way to acceptable performance. As users, we demand the best; we want
the fastest access to resources and faster links to the Internet. We want
our applications to run better, and we want more bandwidth to run multimedia
applications. Cost constraints prevent this from always being possible. In
fact, our requirements as users (playing power-hungry games over the LAN)
are often sacrificed to support business needs. Organisations are not
willing to spend huge amounts of money simply to keep its users happy,
preferring systems that suit business needs, and get the job done. |
