Understanding WAN Bandwidth Delivery
- Introduction to Bandwidth Delivery: How the Computer Crashed into the Telephone
- Analog Modems
- Hierarchy of Dedicated Digital Services
- DS0: The One True Standard
- DS1: The Ever_Popular T1
- The T1 Frame
- Fractional T1
- Fractional T3
- SONET
- ISDN
"We are in great haste to construct a magnetic telegraph from Maine to Texas; but Maine and Texas, it might be, have nothing important to communicate…as if the main object were to talk fast and not to talk sensibly."
—Henry David Thoreau, Walden
Introduction to Bandwidth Delivery: How the Computer Crashed into the Telephone
There has been an electronic bandwidth infrastructure in this country since the 1800s, when Samuel Morse's invention allowed messages to be sent long distances over copper wire nearly instantaneously. Since that time, there has been significant change in the way data is encoded and transmitted. These changes, and the inherent differences between voice and data, have created technology that is often at odds with one or the other. Because voice transmission has long been the goal of the telephone company (or telco), it is the transmission of data that usually suffers.
From the beginning, this communications infrastructure was torn between digital and analog. The initial form of communication over wire, Morse code, was digital in theory. Binary information was designated by short or long pulses (a dot being 0, a dash being 1); however, the transmission media was analog. The pulses were sent as voltage changes along copper line.
Later it was discovered that information could be sent significantly faster in a wholly analog form, and voice transmission over wire took over. After all, who wants to tap out Morse code when you can talk instead? Very fast Morse code operators can send about 30 words a minute—how fast can you talk?
After many people began using voice over wire, the phone company became concerned that it would eventually run out of copper, or places to hang it. A town with 20 telephone owners could well afford 20 copper wires running from individual houses to the telco, but a city of 50,000 or half a million certainly could not.
The goal of transmitting multiple analog signals over a single piece of copper wire became absolutely necessary for telephone expansion to continue. Eventually, some clever person discovered frequency-division multiplexing, which allowed several conversations to be transmitted over a single pair of wires.
How does that work? Well, like many technological miracles, it uses methods that have existed since the beginning of history. In the same way that you can tune in to one voice at a cocktail party and carry on a conversation with that one person while dozens of other people are talking, frequency division multiplexing allows computers to distinguish among multiple conversations going on at once by listening only for signals of a higher or lower frequency than the others.
To a certain extent, that's also the way it's done in frequency-division–multiplexed data signals. Now, you certainly don't want to be in a telephone conversation where you can hear a half dozen other people talking in the background, so the signals have to be separated at the telco before they get to you. But how? It would be extremely difficult for any equipment belonging to the phone company to figure out what voice belongs with which conversation, so the signal is modified before it's bound to other signals. This is done by "painting" the signal to change its characteristics.
For example, you have the following four streams of numbers you want to send along the same wire:
|
6 |
12 |
9 |
3 |
|
8 |
9 |
12 |
21 |
|
6 |
6 |
14 |
9 |
|
17 |
11 |
3 |
7 |
If you just toss them on the wire, it will be impossible to sort them out because the numbers are all very similar. However, if you add 100 to each number in column 2, add 200 to every number in column 3, and add 300 to every number in column 4, you get numbers that are easy to separate at the receiving end:
|
6 |
112 |
209 |
303 |
|
8 |
109 |
212 |
321 |
|
6 |
106 |
214 |
309 |
|
17 |
111 |
203 |
307 |
So, by adding a known quantity to the frequency of the signal, they're made unique. They're decoded at the telco, and the proper conversation is sent to your house along your dedicated phone circuit. Figure 1 shows how it might look as sound waves.
){kind=link}
Figure 1 Frequency-division multiplexing allows multiple signals to travel over a single wire.
Packet-Switched Versus Circuit-Switched Networks
Data transmission can be either packet-switched or circuit-switched. There are advantages to each.
Packet-Switched
Large blocks of data are broken down into smaller parts, called packets. Each of these packets contains all the addressing information to get it from point A to point B. The packets take whatever route is available; there is no fixed path.
Circuit-Switched
A path from point A to point B is negotiated at the beginning of the transmission and kept open until the end of the transmission. During that time, nothing else can use that pipe.
The best example of circuit-switched networking is a telephone call. You dial your best friend, and a circuit is opened between your phone and hers. As long as you have the phones off the hook, that data pipe belongs to you. No one else can use it. If someone calls your house, he'll get a busy signal. A circuit-switched network connection is the fastest and most reliable, but it suffers because no one else can use the line. If you were to take a one-hour phone conversation and remove all the fraction of seconds where there is silence on the line, you'll be left with a lot of silence. Every time you're transmitting silence, you're wasting bandwidth. Unlike people, computers talk only when they have something to say, so their conversations are much more direct—hence, packet-switched.
The Telco Engineers Versus the Network Engineers
When computers crashed into the phone company, there was some carnage. From the outset, telco engineers and computer network engineers were at design odds with one another. After all, they have entirely different goals. The primary goal of the telco is to deliver real-time voice transmission. This can be done at the expense of cost and accuracy. Cost is passed on down the line to the consumer, and accuracy…well, the assumption is that if there's a distracting noise in the background, people will retransmit themselves, usually in the manner of, "Huh? What did you say Bob? There was a bus passing by just then." Network engineers are more concerned that data will be accurate and less concerned that it will be real-time. If faulty wiring causes a handful of static to be thrown into a telephone conversation, most people won't notice, but that same static could alter critical data in the transmission of a computer file, resulting in errors.
What we are left with is a telephone infrastructure designed for analog voice through which digital computer data must be sent. There is a push today for an entirely digital telco that can serve both voice and data. To an extent, this is already happening. Much of the innards of the U.S. telephone system is digital; what remains is known as the "last copper mile," or the local loop of wiring between a consumer's house and the telco building. The investment in that wiring and telephones compatible with it is extensive and expensive, and its replacement will be hard-fought.
For this reason, we have an amalgamation of technologies providing service in sometimes rather strange and kludgy ways.