To many network managers, cabling is someone else's problem. It is, however, impossible to run a network without knowing the nature and layout of the copper or fibre infrastructure that actually carries the data around the company. If you are installing a new network, it is important to know that the investment will hold for a number of years. If you have inherited a network, it is essential to be certain just what speed upgrades the cabling can support.
Imagine, for instance, you have inherited an ageing cabling infrastructure based on Category 3 twisted-pair wiring. This will easily support 10Mbps Ethernet, or 16Mbps Token Ring. It will even run 25Mbps ATM with no problems. Try to install a 100Base-TX switch, though, and strange things will happen, as Cat 3 is not rated to 100Mbps unless you switch to non-mainstream technologies such as 100Base-T4.
In short, you have to understand cabling systems in order to plan for the future and make decisions which you may well be stuck with for up to five years - a perfectly believable lifetime for a cabling infrastructure. Consider that redoing a badly specified cabling system involves not only a sizeable investment, but a huge physical upheaval in terms of actually running the wires under the floors and up the risers.
The most common type of cabling in use today is Category 5 unshielded twisted pair (Cat 5 UTP). Each cable comprises four pairs of core wires, each pair twisted together in an attempt to cancel out internal interference, or 'crosstalk', and minimise the effects of interference from external magnetic fields. Cat 5 is rated to 100Mhz, which means it can support Fast Ethernet and Fast Token Ring (both 100Mbps technologies) as well as 155Mbps ATM.
When the Cat 5 standard was written, it catered for the technologies we just mentioned, each of which uses a single pair of wires for data transmission. As network speeds improved, an obvious way of supporting faster throughput on a 100Mhz-rated cable was to split a higher-speed signal into components and transmit these components over several pairs of cores, reassembling them into the original signal at the far end. Unfortunately, the Cat 5 standard does not guarantee that signals down one end of several pairs of cores will actually reach the far end at the same time. The Cat 5e standard was therefore developed to include, among a handful of other refinements, a constraint on this 'delay skew' effect.
In reality, many Cat 5 implementations do actually conform to the Cat 5e standard. There are no fundamental differences, just tolerances that were not previously specified, but which a well-made Cat 5 system ought to conform to. Delay skew in particular is down to the build quality and raw material consistency of a cable - if delay skew is high, it may well mean that some wire pairs in the cable are of inferior quality, or are inconsistently twisted, thus varying the length (and hence the propagation delay) of the pairs. Verifying compliance (and therefore your network's ability to support technologies such as Gigabit Ethernet) is simply a case of hiring a cabling contractor for a day or two, or even renting a test meter and doing the job yourself.
Cat 5 and Cat 5e are very much today's technologies, but they are now running at the limit of their abilities. With talk of 10Gbps Ethernet being not far over the horizon, the industry is pushing forward to develop cable systems that will support higher and higher transmission speeds. There are two possible paths one can take when planning for a multi-gigabit future: copper and fibre.
Copper cabling has its disadvantages. The main drawback is interference: anything electrical generates electromagnetic (radio) waves which, if sufficiently intense, will interfere with the operation of copper cabling - especially UTP, which has no protection from external interference aside from the limited signal cancellation provided by the twists in the pairs. As the speed of data transmission along a cable increases, each bit of data has to become shorter in duration so that more bits can be passed in a given time - the shorter a data bit, the more likely it is to be eradicated by external interference.
Many network managers simply do not realise the significance of electromagnetic interference on copper cabling, yet in a lab environment we have seen good quality Cat 5 cabling running 155Mbps ATM suffer from significant dropouts when irradiated with interference at a level lower than the background radiation level in a typical office. (Later the same day we banned mobile phones and hand-held radios from the comms room, given that they push out at least 50 per cent more radiation than the average device.)
Interference in the other direction is also relevant to copper cabling. There are limits set by various standards bodies defining the upper limits for radiation leaving data cabling. Since a data signal is effectively just a patterned stream of electricity, and bearing in mind that the higher the data throughput speed the higher the emission is, cabling manufacturers have some work to do to keep cables within their limits.
FTP et al
Given these interference issues, many cable manufacturers have decided to move away from the unshielded aspect of twisted-pair cabling. The twisted-pair approach is still a valid one, as it does an excellent job of cancelling out internal interference, but to apply some kind of shielding to the wire seems to make sense (after all, the co-ax wiring many of us have used for years is known to be almost interference-proof thanks to the earthed outer braid).
The most inexpensive approach is to put a simple foil shield under the cable's outer covering, though some vendors decided to shield each pair from the others. There are various names for these products, though 'foil twisted pair' is a good generic one. Adding this simple shield makes a huge difference to the interference characteristics of a cable; in the same lab test mentioned above, an FTP system coped flawlessly, not only with the basic level of interference, but with 180 per cent of the radiation one would typically find in a heavy industrial environment. So dropping the 'U' of UTP certainly makes a difference and prolongs the life of copper cabling in the network.
This said, copper's days are still numbered. The reason is simple - although interference has been largely dealt with, the actual number of bits one can send down a copper cable is limited, and the 'headroom' between what we are using today and what the cable can take, is minimal.
We have already noted that to run gigabit speeds over standard copper requires the use of multiple wire-pairs within a cable, and to go above 1Gbps will require even more advanced trickery, encoding and compression.
While Cat 6, aimed at providing in excess of 1Gbps in the Lan, is under development (and has been, it feels, forever), it is generally accepted that this will be the end of the line for traditional copper Lans. Although there is talk of Cat 7, it is unlikely to become a reality.
The future is fibre
Why will copper development stop? The answer is the same as with any other shift in technology - the effort required to produce a copper standard that can surpass the proposed Cat 6 specifications would make it an unusable and uneconomical system. In order to recoup the R&D costs, Cat 7 would probably be more expensive than basic fibre, and it would be pushing copper transmission techniques to the limits (the fibre would merely be ticking over). It might even use a completely different physical connector on the cable end, requiring a change in patch systems and an upgrade of network interfaces.
Fibre, especially the less expensive multi-mode cable, is constantly dropping in price, has gigabits of headroom, and is completely immune to all but the most stratospheric levels of electromagnetic interference. By introducing fibre you are instantly unshackled from that automatic assumption of a 100-metre segment limit. There really is no contest. The cost of fibre itself, and of fibre interfaces for network devices, is now little more than their copper based siblings.
The only perceived issue with fibre is that it is in some way complex to understand. While it is true that care must be taken to understand the basics (such as the fact that there are long- and short-wavelength versions of fibre based Gigabit Ethernet, for example, or that there are several variants of multi-mode and single-mode cable), it really is no harder than remembering a few numbers or having a cheat-sheet on the wall.
Not only this, but the traditional problems of joining fibre cables have gone away. Whereas one used to require a skilled (read expensive) cabling contractor to run backbone fibre and do the splicing and connector installation, there are now several pre-terminated products on the market. You tell the vendor what lengths you require, and you receive a reel of cable with a strain-proof cover at each end; once pulled through the riser, the cover is removed and the connector simply clips into a proprietary face-plate.
Fibre in the backbone is now a DIY job, which means that fibre is now available to anyone without the need to spend much more money or to employ specialist contractors.
Of course, it is still true that network speeds may increase over the next few years such that standard multi-mode fibre cannot cope, but some institutions have solved that already by pulling 'dark' single-mode fibre alongside the multi-mode, which can be commissioned as required to take any load that the multi-mode cannot. For most companies, though, multi-mode fibre will be enough for the foreseeable future.
Copper cabling will peak at Cat 6, which will be fine for all but the most enthusiastic power-user desktop applications.
Fibre has shedloads of headroom, is all but immune to electrical interference, and is now inexpensive enough and simple enough to comprehend for it to step smartly in and take over in applications above 1Gbps, where copper starts to pant and wheeze under the strain.
The light at the end of the fibre
Pre-terminated fibre will revolutionise riser connections through its sheer ease of use, and by the time the less expensive multi-mode fibre types begin to run out of steam, single-mode fibre, which has a greater capacity and longer length limits, and which is therefore currently rather more costly, will be far more mainstream and hence more sensibly priced.
Most importantly, the boundaries of fibre development are constantly being pushed. BT, for example, have run Gigabit Ethernet over many tens of kilometres in the lab, and with up-coming technologies such as wave division multiplexing, the physical bandwidth of fibre cable is also expanding.
Facebook told by Brussels-based court to stop tracking non-users and to delete all data held on them
Supply chain and manufacturing experience could give Dyson an important edge
New VR Zone Portal arcades open in London and Tunbridge Wells
Systems-on-a-chip with integrated AI features could make voice and facial recognition