Local area networks

Press, 2 April 1985, Page 36

Local area networks

Computer networks are things of wonder and beauty. No-one agrees exactly what they should be. how they should work, and of what they should be built. IBM happily plays a cosy game of suggesting what parts of the network should be — but so far has carefully omitted to give the full details of what would, almost overnight, become the de-facto standard of the industry except for a product called PC Net which is a low-end standard. Where full-scale personal computer networks are concerned, your guess as to what the eventual standards will be is as good as the next person’s — and very possibly much better. The basis of any true network is to link a series of computers so that they can communicate, one with the other, after a fashion. A local-area network, which the acronym-mad computer industry insists on referring to as a LAN, does precisely the same thing on a limited basis within a specific geographical area, normally and ideally a single building.

The advantage of such a network is that fairly expensive equipment can be shared between several work-stations. For example, instead of every computer

having its own laser printer, three, four or more can share.

A further advantage of linking a group of personal computers into a local-area network is that electronic messaging becomes extremely simple and effective. There is no doubt whatsoever that electronic messaging reduces the use of the übiquitous and timeconsuming memo and, after the staff get used to it, becomes a very functional and useful business tool.

If all were for the best in the best of all possible worlds all computers could , freely communicate with one another.

Sadly, this is a far from perfect world, and you need a fair amount of compatibility between machines that you hang on to an area network or you will get strange errors and glitches which will upset the harmony of the whole. A true local-area network will normally work on a

ring system. The classic example is Ethernet, which very nearly became an industry standard but is now starting to show its age and seems unlikely to make it as the de-facto standard. Working on the same basis is the Corvus Omninet which is widely installed throughout Australia and now appears to have overcome some early teething troubles. Other local-area networks are built around a star configuration with big brother in the middle controlling all the satellites through some form of multiplexer system. A local-area network normally tries to send information at the highest possible speed compatible with accuracy. If the system is extensive and the user is to feel that there are no hiccups, no wait times, no queuing, then those speeds have to be fairly quick. With information frequently travelling at speeds around the one megabit

mark, protocols have to be established to stop errors creeping into the system and to make sure that the correct information goes to the correct destination.

These protocols set up the way in which the information will be sorted out into packages and the way in which those packages will be addressed. They also lay down the exact recovery procedures to be used automatically if the information gets corrupted in transit.

Almost all microcomputers are suited to work ■as part — normally called a node — of a localarea network. In most cases they are fitted with a port for serial transmission, and again, in the majority of cases, this is an RS-232C plug which follows a set international standard.

Using these ports it is usually relatively simple to set up a local-area network. Always providing that the machines connected to the network are all compatible. Where the major problems occur with a localarea network is when it ceases to stay within the true meaning of the term and gets extended and stretched into something else.

The simplest example is when a local-area network

is asked to join together machines which are in different buildings. The only way this can be done with any great assurance of success is to physically link them together with wire or fibre optics. This can create great trauma and expense if the connection has to go under a road owned by the local council. The costs of inter-connecting then become stratospheric. Another problem that occurs, frequently within universities, is that a local-

area network is stretched to its absolute theoretical limits, simply because the demand for interconnection is always there, and always growing. There, is as yet, no simple solution to this problem. It is just theoretically possible that radio transmission could step in, and much experimental work has been done in this area, specifically in Canada. But it seems unlikely this will be a viable and workable alternative in the short-to-medium term.