Structured wiring is a rationalised type of data wiring, usable for many different purposes. It's also known as Premises Distribution wiring, and by a number of different trade names: ours is called OSCA by BT. To get an idea what it is, imagine a primitive manual telephone exchange. Refer to the diagram.
[If your browser permits it, you may wish to open this diagram in a separate window and keep it in view as you read the following text.] Each subscriber's socket has been permanently wired to a corresponding socket on a big switchboard: these wires are depicted as thick. straight lines on the diagram. To get service, they have to contact the switchboard operator, and ask to be connected to the service that they require (these "patch" connections are depicted by the thin, curved lines on the diagram). In a large system, the 'switchboards' would additionally be interconnected with highspeed data highways, such as fibre-optic links, but (at least for the present), our department does not need that. The specification of the wiring that's used is much superior to ordinary telephone wire. The cable contains four twisted pairs: one cable is used per individual 8-pin socket. The question of which wire goes to which pin, and in particular which pin is paired with which, is laid down in a standard. From each room socket, this fixed cable leads to a corresponding socket at the 'Patch Centre' (which is our manual switchboard). To make use of a socket, for a particular purpose, you request us to connect it to the data service that you want. Normally, only one service is connected per socket, at any given time. We will see in the next segment what we mean here by a 'data service' (e.g the Departmental Ethernet): do not confuse it with the actual destinations (other computers) with which you want to communicate. Your connection to a particular service is likely to stay for months or years (unlike a phone call!), so my telephone analogy is a bit weak. The connection is made at the patch centre by inserting a so-called 'Patch Cord' between 'your' patch socket and the desired service.
In our building (the overall dimensions of which are about 82m x 53m and on 4 to 6 levels), there are in fact two patch centres, one on level 2 on the old side of the building, and one at level 4 on the new side. The design calls for every socket to be not more than 90m of cable from its hub. The two hubs are interlinked by an ample number of fixed cables of the same type (as well as by a couple of coaxial cables not shown). In the typical telephone installation, the individual subscriber tails would be patched into large multipair highway cables, which in turn are patched at distribution boards around the network: there are numerous joins, and many wire pairs share a common sheath for much of their run. The data wiring, on the other hand, has no joins between room socket and patch centre, and each socket uses an individually sheathed cable, of the type already mentioned. This preserves the transmission-line characteristics for high-speed data, and reduces the chances of interference between circuits. But there is no shielding in the cable, i.e it is UTP (unshielded twisted pair): interference is minimised by using a tight and uniform twist. STP (shielded twisted pair) cable is much clumsier and more expensive (to purchase and to install), suffers higher signal losses, and is rarely necessary in our kind of environment as a protection against interference (it may be useful against electromagnetic surveillance, but fibre would be a better choice against both).
Some experts swear by STP, and decry UTP; normal IBM token ring gear expects STP, although a UTP variant is available. For all new data wiring, we are using the new UTP structured wiring, and it has proved excellent. Anyway, the networking protocols include mechanisms for detecting and recovering from corrupted data, within reason.
The diagram shows a number of room sockets patched to two of the various services available at level 2; please imagine several other services there, and the same at level 4. It also shows an example of linking one room socket to another, which is another usage that's occasionally convenient. Finally we see a socket at level 4 patched through one of the inter-hub link cables to a service available at level 2 (the converse is of course also possible). However, this trick might not always work if it exceeds the permitted wiring length for the service in question.
Having seen a neat diagram, it may be instructive to look at the real thing.
Here is the patch centre at level 2 on the old side of the building. It consists of two standard racks. BT's neat patch panels are almost hidden behind a tangled mass of patch cords. At the upper left we see the original Ethernet hub (24 ports), and at the upper right is the CAMTEC X25 PAD. Low on the right hand side, with patch fields above and below them, are some 3COM Ethernet hub modules. Some other gear is either hidden from view or too inconspicuous to point out here.
It would be technically feasible to run telephone circuits over our data wiring, but we decided not to do it. British telephone regulations have been rather strict about what equipment can be attached to wiring that also carries telephone circuits, and we did not want to get involved in that. The standard connector for data wiring is the large American phone jack, type RJ45, or its smaller brother the RJ11; the British telephone socket was specifically designed to be incompatible with these, so we are running no risk of accidental mixups.