What is the number one thing that adds to the cost of running an airplane?
Why go and reinvent the wheel in regards to ethernet and Cat5?
Cat5 is wasteful. Ethernet needs 2 pairs of wires, or four cords total. Cat5 has six. Why? Corporate convenience. You are wiring an office building, for most corporate wiring jobs, your costs are in labor and time. Weight (below a sane threshold) has zero bearing on cost, except maybe for shipment. It is nice to have some overhead to run future services over, such as digital telephone systems or Power over Ethernet (PoE).
Airlines are the exact opposite. You install the wiring once, and you pay for it each and every time the plane takes off.
So companies come up with various standards for reducing this weight. When you have literally tens of thousands of miles of cable in an airplane, reducing the wire weight by 1/3rd can make a big difference!
Then they start fighting over whose standard is better. Boeing used to prefer a way that was a single pair put together. Almost no cross talk, interference is dropped to nill, really cool advantages. It can do some really long wire runs. The disadvantage is that you need to use 2 of these for each ethernet drop, since as said above, ethernet needs 2 pair.
Airbus pushed a standard that uses 4 pairs. It also has some cool electrical specs on it, though not quite as good as the Boeing stuff.
Both standards blow the pants off of CAT5 though, in every respect possible.
Of course the standard RJ45 connector is no longer any good, not to mention that it never held up to (nor was it designed to!) vibration and stress requirements. So you have these connectors that cost $5 or more apiece, and are huge and made out of steel. I couldn’t figure it out at first, until the first time a tech document mentioned vibration, then it hit me. Having your plane’s computers vibrate themselves off of your LAN is not a good thing!
These end connectors are so well measured and defined, that the increase in signal loss that will occur each time you plug and unplug the connector is well defined.
It is not that RJ45 is unreliable per say, but we have all had RJ45 cables where the little plastic snap thingy falls off and the cord subsequently tends to fall out. For consumer stuff, this is not a problem. Unplug the cord, cut off the current RJ45 connector, and crimp on another one. If you don’t know how, you can learn in just a few minutes, it very easy unless you are color blind. (Not a joke CAT5 cords are color coded, each ethernet cord has 12 small wires inside of it, each in a colored plastic cover.)
On an airplane, it can be significantly harder to do this given that:
- You have many magnitudes more wires with a far simplier wiring layout, so isolating the fault is not as simple.
- You are flying in an airplane!
Doing network debugging and troubleshooting at 10k feet does not seem like a good idea to me.
Oh, and of course, everything in airplanes is redundent. Redundant nics, redundant wires, the entire network is redundant.
You want to know how paranoid the airline industry is?
They looked at ethernet, went “Hey, that is a good idea. But, umm, we’re going to have two plugs on each device, and each device will be hooked in to two different routers. Ah, that’s better.”
That is paranoid folks.
Not unfairly so, everyone working around networks has also likely encountered a switch or a hub that for some weird reason, has dead ports on it.
I also figured out why the MTBF listed for a given part electronic part on a plane tends to be 2/3rds the industry standard for that part.
Once again, vibration.
Pick up the monitor in front of you. Now shake the living crud out of it. Keep doing that for a bit. Good. Now repeat it every four or five hours. I am sure you can imagine that if you keep it up, you are going to reduce the lifespan of your monitor. 🙂
 Mean Time Between Failure. A measure in hours of how long until a part, on average, will die. Given because you want to know how long an investment in a given part will last so you can amortize it over time, and also so you can give a best guess estimate as to when to replace the part before it does fail, in the case of mission critical applications.