I recently had cause to do some research into ethernet wiring standards as part of a troubleshooting exercise. A quick trip to the Google yielded all manner of information of varying quality passed off as absolute fact. Some of that information is clearly nonsense and I wanted to address some of that. First, I’ll start with some background and then move on with the wiring bits.
We don’t need to go into gory details about how ethernet functions for this discussion. There is one critical aspect that is important. The rest is just signalling and implementation details. In particular, I will be concentrating on metal wire (copper, etc.) even though there are other media that behave like ethernet, such as fibre optics.
Modern ethernet typically runs over what is known as “Unshielded Twisted Pair” (UTP) cabling. That is, it runs over pairs of wires that are twisted together with no extra shielding around it. This is opposed to shielded twisted pair which has, get this, shielding. The shielding might be beneficial in environments with a lot of electromagnetic interference, but the extra cost of the shielding isn’t required in most cases. There are also coaxial cables which have a centre conductor and a shield separated by insulation. The original ethernet used coaxial cabling, but you are unlikely to encounter it in any recent wiring job and you wouldn’t be reading about ethernet wiring standards if that was the case. UTP is one of those cases where the benefits over the previous coaxial standards were so significant that UTP basically took over the world (and, consequently, made a few other things easier).
Anyway, UTP (and its shielded cousin) provide what is known as a balanced transmission line, as long as two wires in the same pair are used. There is a bunch of physics involved, but the density of twists along the length of a twisted pair along with the wire gauge, conductor quality, and insulation combine to allow the ethernet signal to run from one device to another. Different parameters have different bandwidth limits which is why there are different “categories” of UTP with “Category 5e” and “Category 6” typically being used today. Faster ethernet speeds require higher bandwidth cables.
A complete UTP ethernet connection uses two pairs, one for transmission and one for reception, for speeds of 100Mbps and lower. (Gigabit speeds use four pairs). This allows for both transmission and reception to occur simultaneously without interfering with each other and without some form of multiplexing modulation. Historically, UTP connections went from an end device such as your computer to a hub or switch. The difference between end devices and switches is which of the pairs is connected to the receiver and which to the transmitter. As a result, if two devices of the same type (two end devices or two switches) are connected directly, a crossover cable is required to make sure the transmissions from one device reach the receiver on the other. (Modern devices mostly have automatic detection for whether to swap the pairs or not so in most cases, you can get away with regular straight through cables.
There is one final note about the typical cabling used. Most ethernet cables use a four pair UTP cable which means there are four twisted pairs within a single jacket. These pairs are fed into an RJ45 connector. This is where the wiring standard comes in. Pair #1 goes to the middle two pins, pair #2 goes to the pins on either side of that, pair #3 goes to the left most two pins and pair #4 goes to the right most two pins when looking at the connector with the retention clip pointed away from you. The wires for each pair must go to the same pin on the connector on either end of the cable.
There are two standard ways of wiring an RJ45 connector for ethernet. These are listed in the table below. Each pair will typically have a wire with a solid colour and one that is white with coloured stripes. I will indicate the striped wire as “white-colour” and the solid wire as just “colour”. The pin numbers on the connector go from 1 on the far left to 8 on the far right when looking at the connector with the retention clip away from you. The two standards are “T568A” and “T568B”.
You will note that the only difference between the two is that the green and orange pairs are swapped. I have numbered the pairs in the table above reflecting the description above in the background section. There is a historical reason for this ordering that is unimportant for this discussion.
The critical point is that each twisted pair is wired up to one of the connector pairs, and that the order of the white and coloured wires in each pair is the same at either end of the cable. It makes absolutely no difference which colour pair connects each connector pair within a single cable. A cable could use brown for pair #1 and blue for pair #4 and it would be perfectly functional. All pairs within a UTP cable are electrically identical, or at least they are supposed to be.
If it makes no difference which pair connects each connector pair, then why do we need wiring standards? This is for a simple reason, actually. If multiple people are working on the wiring for a network, it is far less likely that problems will arise if everyone is using the same wiring standard. While it is trivial to make sure both ends of a single six foot patch cable are wired identically, this is not so easily verified for a two hundred foot cable run between floors in a building. Thus, regardless whether you use T568A or T568B or something you made up yourself (that keeps the pairs matched properly), you need to document it and stick with it so future people working on the network can figure out what’s going on. However, for individual patch cables, there is no need to stress about which order the pairs connect as long as they are wired properly pair-wise end to end.
Note that you can create a crossover cable as described in the background section above by using the T568A standard at one end and the T568B standard at the other end. Normally, however, you would use the same standard at both ends to create a straight through cable.
T568B is better for the signal. This is a load of crap, period. There is no functional difference between the pairs in a UTP cable. Any pair is completely interchangeable with any other pair. One source (which I will not link because it does not deserve further exposure) claims that “T568B is designed for better signal isolation and noise protection for newer networking systems and products.” There is no possible way to obtain “better signal isolation” or “noise protection” by simply swapping the green and orange pairs. The signals are still on the same pins on the connectors, which are still connected through the balanced transmission line of the pairs.
You can’t mix T568A and T568B cables. This is also nonsense. As long as you don’t mix the two types within a single cable, you can freely use both T568A and T568B cables within the same network. The electrical signals can’t see the colour of the wire insulation. As long as each pair is properly connected end to end, it makes no difference what colour pair is used at any step along the way. The signal can run along a blue pair from the server to the wall plug, a green pair from the wall plug to the wiring closet patch panel, and an orange pair from the patch panel to the switch without having any problems. As long as your cables are wired properly for whichever standard they are using, they are fully interchangable.
To sum up, it makes no difference whether you use the T568A or T568B standards for making your patch cables. Just don’t mix standars within a single cable. The same applies to in-wall wiring, but it’s definitely better to pick a standard an stick with it for that to prevent accidental mixups.
Further, neither standard is technically better or technically inferior. The colour of the insulation on the wire makes absolutely no difference as far as its electrical properties are concerned.