I wouldn't call 30Mbps a connection speed. Speed is related to latency - how fast a packet can travel from point A to point B. While bandwidth refers to how much can come through in a defined period of time.
If you compared it to plumbing, speed is how fast the sewage flows through the sewer, while bandwidth refers to how big the sewer pipe is.
So bandwidth in this example would be related to the bandwidth of the RF signal that's allocated to the "channel" that's carrying the downstream signal the data is being carried on. Connection speed is going to be dependent upon the physical characteristics of the medium, and the ability of any switches or routers to hand-off the packets efficiently.
However, most people commonly think of speed in terms of how much data they can get in a certain period of time. In that sense, speed would be more related to the width of the pipe, as opposed to the speed at which the crap flows through the pipe. In other words, unless latency is
*very* high, people notice how much stuff flows out the end of the pipe, and have no idea how fast it traveled from one end of the pipe to another.
You're really not as confused as you think. If you understand bandwidth from an analog perspective, you understand it from a digital perspective as well. Common usage of terms such as bandwidth and connection speed isn't always accurate.
Lookup "Shannon's Law." It is the theory that provides the maximum information rate for a channel given the channel's analog bandwidth and the SNR of the channel.
Of course, the implemented data rate of a channel is determined by the modulation method and error correction codes selected by the system designer. These issues are a trade off of computational power and cost to produce a system that is appropriate for the application. However, no system can exceed the Shannon limit for it's channel.
Cable modems in the US use a 6MHz wide channel with 64QAM or 256QAM modulation, at a symbol rate of up to 5 MSymbols/sec (i.e. a little less than the 6MHz bandwidth). At 5Msym/sec with 256QAM (8 bits/sym), the raw data rate is 40Mbps. Error correction lowers this to about 38Mbps. 64QAM produces the 30Mbps number you found.
For a great tutorial on DOCSIS and the modulation issues around it, read the three part series written by Rob Howald in CommDesign. Start here:
There are two types of bandwidth of interest here. One is the data rate, measured in bits per second, which is what you're buying from the cable company. Then there's the channel bandwidth, which corresponds to the frequency spectrum required to carry that data. The signalling rate, measured in baud, must be less than the channel bandwidth. The two types of bandwidth may be, but not necessarily are, related. In order to get a higher data rate, into a channel that you'd think wouldn't be wide enough to carry the data, multilevel encoding is used. With this, each signalling element, over the cable, can carry multiple bits of the data. This is similar to what happens with telephone modems, which can carry up to 54 Kb s over a 3 KHz voice channel. Look up Shannon and Nyquist, for the engineering basics on this. There are also many excellent communications technology references, which cover this.
How nice of all you to respond in detail to my question. I appreciated your reference to Shannon Law and Nyquist. I need to dust off my old engineering books. Also, your links to Rob Howald are excellent and I will review that. And, justifying the magazine article of 30Mbps was appreciated.
To make matters sillier, popular jargon now uses "broadband" to mean "high data transmission rate" and applies the term even to baseband data networks. And "narrowband" is now being used as a label for dial-up connections via old-fashioned telephone modems. For example, EarthLink now refers to its dial-up users as "narrowband subscribers".
I suppose the next speed increase will be get a moniker like "ultraband"; although the marketing types push "lightspeed" for optical fiber. I can hear it now: "Why use pokey old electrons when you can surf the web on a beam of light?"