16

I'm reading Computer Networks - A Systems Approach 5th ed., and I came across the following statistics for the speed of light through different mediums:

Copper – 2.3 × 108 m/s

Fiber – 2.0 × 108 m/s

So, are these figures wrong, or is there another reason to explain why copper is worse than fiber? Does fiber have better bandwidth (per volume) or something?

38

No, the numbers are right (Page 46). If I can reword your question, it's "Why should I use fiber if the propagation delay is worse than copper?" You are assuming that propagation delay is an important characteristic. In fact (as you'll see a few pages later), it rarely is.

Fiber has three characteristics that make it superior to copper in many (but not all) scenarios.

  1. Higher bandwidth. Because fiber uses light, it can be modulated at a much higher frequency than electrical signals on copper wire, giving you a much higher bandwidth. Also the maximum modulation frequency on copper wire is highly dependent on the length -- inductance and capacitance increase with length, reducing the maximum modulation frequency.

  2. Longer distance. Light over fiber can travel tens of kilometers with little attenuation, which makes it ideal for long distance connections.

  3. Less interference. Because fiber uses light, it is impervious to electromagnetic interference. That makes it best for "noisy" electromagnetic environments. Also, fiber does not conduct electricity, so it can electrically isolate devices.

But fiber has drawbacks too.

  1. Expense. The optical transmitters and receivers can be expensive ($100's) and have more stringent environmental requirements than copper wire.
  2. Fiber optic cable is more fragile than wire. If you bend it too sharply, it will fracture. Copper wire is much more tolerant of movement and bending.

  3. Difficult to terminate. Placing a connector on a optical fiber strand requires precision tools, technique, and expertise. Fiber cables are usually terminated by trained specialists. In comparison, you can terminate a copper cable in seconds with little or no training.

  • 11
    Great answer, I'd like to add one thing if I may. In large scale environments that require tons of fiber, it's physical profile is much smaller, and it can handle much higher cabling density as a result. – Jordan Head Feb 1 '15 at 16:19
  • @Jordan, that is true. We have further reduced the cable density by using bi-di SFPs which uses a single single mode fiber strand for 10G vertical links. Yamasaki made them somewhat affordable... – user4565 Feb 2 '15 at 20:44
  • I like this answer too. I'm not sure about the expense - I think the answer applies more to "local" than it would to "long distance". Laying enough copper at the bottom of the ocean to reliably carry even 1 Tb/sec from one continent to the other might not be so cheap. In fact I'm not even sure what that would look like! – uhoh Jan 10 '16 at 11:07
  • 3
    @InterLinked "analog signals are superior in terms of quality to digital" - got a source? – user253751 Feb 21 '18 at 21:47
  • 1
    @InterLinked Lower latency I can understand, but digital signals are inherently immune to noise (but if there is too much noise, they drop entirely, instead of letting you still have a very bad signal). But then they also cheap out by using 8bps 8kHz resolution which may be much lower than the effective resolution of your analog signal. So your specific analog phone signal may be better quality than your specific digital phone signal, but in the general case, you cannot say analog signals are better. Usually the noise immunity wins out overall and digital is better. – user253751 Feb 25 '18 at 7:44
6

I would like to add one benefit with fiber connections. Consider a connection between two buildings with different ground potential. If you where to use copper in this situation you could end up with current leakage and possibly a dangerous situation. This is not the case with fiber because it isn't a conductor.

  • 1
    Only if you use a non-metallic fibre cable, which is a bit of a speciality cable. Most burial cables have some sort of metallic reinforcement, which has to be grounded at both ends for safety reasons (induced voltage and such, we calculated it to around 6kV on a fibre cable running along an underground 20kV power line for around 10 km). Still, better to have the fibre cable grounded to the ground busbar than the shield of the CAT cable grounding to the port on the network switch. – Stuggi Jul 2 '18 at 9:35
4

The propagation speed is often expressed as the velocity factor of a medium - the fraction of the speed of light you get.

On the physical side, light going through a medium is slowed by the medium depending on its refractive index. Fiber has the added 'problem' that the core requires a slightly higher refractive index (optical density) than the cladding to properly guide the wave. The effective propagation speed is the speed of light divided by the refractive index, or the velocity factor is the reciprocal of the refractive index. Most fibers have a velocity factor of or close to .67.

Copper is a bit more complicated. The actual electrons are not moving substantially, it's rather an electric wave (field fluctuation) flowing through the cable - somewhat comparable to sound in air. The propagation speed of this wave surprisingly does not depend on the conductor alone but on the combination of the conductor and especially the insulator (its permittivity) because the wave needs to propagate through the latter as well. The effective propagation speed is the speed of light divided by the square root of the permittivity.

For copper, a velocity factor of close to 1.00 is possible by using air as insulation as with special coax cables or open ladder cables. Copper network cables range from .77 (RG-8 for ancient 10BASE5) to .585 (Cat-3 for 10BASE-T) with the common Cat-5e and Cat-6 at .65 (=slower than fiber).

As has been pointed out, in practice, there are lot of other factors contributing to the effective propagation delay such as transceiver technology, encoding overhead, forward error correction and possibly retransmissions. The velocity factor isn't usually critical.

As to fiber "being better" - it's got higher performance for sure, but "better" depends on your requirements, including cost.

1

I believe another reason fiber propagates "slower" than copper is because the light is, by definition, refracting across the fiber along the distance of the cable. The physics stack exchange has a different take on this:

https://physics.stackexchange.com/questions/80043/how-fast-does-light-travel-through-a-fibre-optic-cable

  • 2
    Not true - in single-mode fiber, there's only straight propagation with a single mode. The medium literally slows the wave - this is the exact reason for refraction as well. – Zac67 Feb 22 '18 at 11:57
0

It depends on distance and number of transactions/users

Copper is better than fiber over short distances (under 10 meters) where bandwidth requirements are currently under 40 Gbit per second. At higher rates and distances the packet loss rate rises to over 50% quite rapidly. To correct this requires repeaters which rapidly raise the latency, and cost of connections.

Even a 10% loss will cause at least 1% of end users to perceive up to 10x latency increase.

Fiber is better than copper where the bandwidth is over 100G and distance over 1 Kilometer and number of users per network exceeds 1,000.

Both Fiber and Copper have massively lower cost expansion capability than wireless, but for many of the domains between copper and fiber, wireless is a rapidly deploy able, unreliable, ever degrading connection medium.

Any mechanism that can be used to expand wireless bandwidth, can almost always be used to immediately expand copper and fiber bandwidth.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.