|
This chapter has discussed
some of the practical aspects of networking. As someone who has
run a network for many years, I have found that the ping and traceroute
commands are invaluable for tracing faults. One network that I used
to look after was a 10BASE2-based Ethernet network. Unfortunately,
as it was based on coaxial cable, whenever someone disconnected
their computer, it would bring the whole network segment down, or
it would slow it down until it was hardly useable. One method that
we used to bring the network back up was to disconnect all the computers
from the segment. A node nearest the correctly operating segment
would be connected, and the other terminated with a terminator.
The node was then tested to see if it could log into the network.
The nodes were then put back on-line one a time, until the fault
was traced (as illustrated in Figure 1).
A Digital Voltmeter
(DVM) was a useful as it could be used to test for the heartbeat
voltage. In these days of network hubs and switches, the best method
of determining if a node is operating is to view the green LED on
the network card and the hub/switch.
The great problem
that I had was when I looked-after a Token Ring network, which worked
well, and was very efficient. Unfortunately, a network fault caused
terrible problems, as the whole network would go down. To improve
fault detection we disconnected each node in turn, and inserted
a by-pass cable. The problem from there was getting the ring back
on-line, and for it to generate a new token. As the network grew
in size, the more difficult it became to find faults. Thus, Token
Ring is an excellent networking topology but, at the time, suffered
from too many problems. The way was thus clear for Ethernet to carve
a massive niche for LANs.
One method
to reduce the number of nodes which are affected by a fault is to
segment the network, using bridges, switches or routers. A router
offers the best segmentation as it intelligently routes data in
to and out of a network segment. Routers have many advantages over
bridges, as they do not forward broadcasts (which are typically
used with ARP for nodes to find-out the MAC address for a given
IP address), and not to forward traffic with an unknown address
(as a router will only forward is they know that the destination
is outside the network segment).
Figure 1 Fault-finding
on a bus network
© W.
Buchanan, 2000
|
