By Jason Romney and Edward Jozis
Computer networks come in many shapes. Jason Romney and Edward Jozis report.
The kind of computer network in which each connected computer has equal access to the resources of every other computer is called `peer to peer'. The advantages of this kind of network have been described here recently, but it is certainly not the only type of network available.
Although a peer to peer Ethernet network is possibly the best kind of network for use in homes and small businesses, it is unquestionably cheaper to simply hook together two computers through a program such as Brooklyn Bridge, the shareware program $25 Network, or Microsoft's DOS 6 file transfer utility.
This kind of connection is made via the serial (or parallel) ports at the rear of the computers to be linked and transfer rate is about the speed of a floppy disk. Such a network solution offers slow but reliable data transfers and sharing and it is a good taste of networking.
Another alternative to a peer to peer network is the most common kind: a file server network.
The main reason for creating a computer network in your business is usually to share data. For example, you may need to look at the quotation written by someone in another part of your organisation or, if you have two people writing invoices, invoice numbers must run consecutively.
One solution is to simply throw a floppy disk across the room or yell out what invoice number was used. These solutions may exhibit a certain rakish charm, but definitely fall down in the corporate image department.
Another choice is a central processing computer. Geographically separate people will each work at computer terminals connected to the central computer via cables. These terminals may merely be dumb ie with no hard disk or Central Processing Unit (chip), and rely on the central computer for almost everything. In that case, as more people launch processor intensive tasks, the processing slows down.
A similar, but now more common solution, is where information will be stored in the hard disk of a central computer known as the file server. All the workstations linked to that server will be able to access a single copy of a centrally stored and shared version of the information.
Unlike in the mainframe model, with a file server arrangement processing power is spread amongst your users instead of centralised. Thus, the network doesn't slow down as much just because someone is doing a particularly processor intensive job.
Indeed, on a small network the file server doesn't have to be a high powered 486. Even a 386 25MHz provides quite acceptable performance on a network which is not very disk intensive. The main thing about the file server is that it needs a fairly large hard disk eg 400 MB - although its main purpose is not to increase available storage capacity, but rather to centralise files and share them in an orderly manner.
When, for example, stock or inventory information is contained on a centralised server, duplication and possible data inconsistency are avoided and implementing data security becomes easier. Of course, on the other hand, it might become easier to steal the data from a remote site and, if the central server crashes, everyone may lose everything.
The primary function of a file server is to send files to users. It doesn't do any real processing at all.
Usually the work stations are PCs, connected via coaxial cable (similar to that which runs from antennae to your TV) or twisted pair telephone-type cable which provide the 10MB per second Ethernet data transmission standard.
In addition, there are ways of connecting to all networks via a modem from a remote location such as your home.
It is worth noting that with a file server network, if a workstation does a big database search all the data is transmitted across the network and other people who want to get a file from the centralised file server get slowed down.
The client/server configuration is a variation on the file server theme in which a computer physically close to the server performs the searching of the information and transfers only the wanted information across the network. Thus network traffic is reduced and overall network performance is undegraded.
Not many software applications can provide this, but examples are Oracle, Ingres (a database system) and Microsoft Access (acting as a front end to the Microsoft SQL server, another client/server based database application).
So these are some of the ways you can connect your computers. As we have seen in the last weeks, for domestic use, a peer to peer network is probably best because resources can be shared. The disadvantage, if any, is that such a network may slow down when each workstation, apart from doing its own work, has to serve files to other people on the network and give access to a local printer if a person somewhere else wants to use it.
For businesses, a file-server model aims to ensure data consistency so that every part of the business knows about things done by the other parts. On the other hand, the file server model has the disadvantage that you need to set aside a whole separate computer to be the server, which adds to the cost of the network.
Good luck.