RE: ABC news on Internet Telephony

At 12:37 AM 7/21/96 +0200, Remo Pini wrote:
A single central office has many times the bandwidth of the widest part of the internet, and the average state has hundreds of CO's. If even a small portion of the Internets current users tried placing a call things would grind to a halt. A huge increase in the number of backbones and their bandwidth would solve this, but who will pay the bill?
I guess Internet-telephony is one of the bandwidth killers.
Potentially. However, there has been some mention of a new standard for voice compression that puts voice into 2400 bits per second, a factor of about 25 lower than the phone company normally uses. (They use 8,000 samples per second at 8 bits per sample, companded.) At that rate, a pair of modern, 2.4 Gb/s fibers could handle 1 million simultaneous phone calls. Since some of the newer fiber systems put 8 or more separate channels down a single fiber, that would work out to 8 million conversations. I have to conclude that we shouldn't even be close to running out of Internet capacity, _IF_ it were driven by state-of-the-art fiber and similar-speed switches. But it probably isn't. At best, Internet probably only gets a fraction of the capacity of a given fiber wherever it flows. This will have to change.
TANSTAAFL
Sometime ago the discussion was on the cost of laying new fiber, may I suggest the realworld heuristic of "a million dollars a mile."
There are of course a lot of alternatives:
In most cases, "new fiber" isn't needed, and will probably only be rarely needed on long-distance links. As I understand it, most cableways are laid with extra tubes, into which new fiber cables can be blown in (using compressed air) long after the trench is filled. The specific example I saw, there were three 2" diameter tubes in a larger tube, and according to the contractor (I asked...) only one of the tubes would be filled at that time. In addition, while he wasn't sure, he thought that at least some of the 36-fiber cable in that one tube would remain "dark," or unused until it was later needed. I don't know how expensive it is to add that extra fiber cable into an existing tube, but it would be VASTLY cheaper than the original trenching operation. Further, much of the improved transmission technology can be used on the older fibers to increase their capacity: A fiber now used to transmit a single 2.4 gigabit signal can be upgraded, simply using new channelized transmitters and receivers to increase the data rate to 8 or 16 times the previous rate. Jim Bell jimbell@pacifier.com

The internet can't get that much capacity, we don't have swiutching technology beyond the test phase to handle gigabits of data per second, and we don't have the routing technology to move packets from point A to points B-ZZZ when searching through a routing table hundreds of thousands of lines long that never has a chance to stabalize between the change messages that keep coming in. If you're interested, search for the writings of Noel Chiappa, who talks about this regularly on ietf, big-internet, etc. Adam jim bell wrote: | | Potentially. However, there has been some mention of a new standard for | voice compression that puts voice into 2400 bits per second, a factor of | about 25 lower than the phone company normally uses. (They use 8,000 samples | per second at 8 bits per sample, companded.) At that rate, a pair of | modern, 2.4 Gb/s fibers could handle 1 million simultaneous phone calls. | Since some of the newer fiber systems put 8 or more separate channels down a | single fiber, that would work out to 8 million conversations. | | I have to conclude that we shouldn't even be close to running out of | Internet capacity, _IF_ it were driven by state-of-the-art fiber and | similar-speed switches. But it probably isn't. At best, Internet probably | only gets a fraction of the capacity of a given fiber wherever it flows. | This will have to change. -- "It is seldom that liberty of any kind is lost all at once." -Hume
participants (2)
-
Adam Shostack
-
jim bell