Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Linux Links | Bling | About BLU |
Jerry wrote: > It would be nice to see some people actually start using DSL services. Where > DSL is a guaranteed bandwidth to the provider, cable modem is shared over an > effective LAN. I have yet to experience any slow downs that I could > attribute to traffic within M1Xs network. And in both cases, once you get up > to the provider's sites, you are in a shared bandwidth. M1X limits the > number of cable modem subscribers on a segment. I was told that if that > number is exceeded, they will split the segment. As a fairly major developer of xDSL service, I concur with this. The limitation of the rival technology, cable modem, is not in the shared segment. AT&T crafted a specification, which apparently most other companies are adopting, which calls for exactly 420 "passed homes" on each cable segment. The DOCSIS standard calls for 40 megabits of raw capacity on each cable segment. Doing the math, if 100% of your closest 419 neighbors are (a) subscribing to the 'net, and (b) downloading at the same time you are, your share of download capacity is 100 kilobits--faster than today's modems. And the odds of everyone using it simultaneously are, well, precisely zero unless you conference everyone together and do a deliberate test. The head-end hardware and the monitoring systems are where the bottlenecks are. If M1X or RCN or whomever isn't aware of the usage levels on a segment, they won't bother splitting it; if they don't have a way for customer service to give feedback to the engineers who build cable segments, they won't upgrading head end capacity. All that sits at a cable head-end for Internet service is the cable-modem head-end devices for each segment served, and some Cisco router hardware to feed data back to a transit pipe (to companies like UUNET or GTE Internetworking or perhaps a NAP if the cable company is a member of one). The Cisco router hardware works the same at RCN or Shore.Net or wherever. The size of the pipes to each head-end site can range from a single T1 to multiple OC-3's; if I had to guess, most cable providers use a T-3. The monitoring software and procedures differ from one company to the next, which means they may not keep close track of whether the head-end site pipes are full. That's where you'll find the quality differences. Those quality differences will occur across ISP's regardless of how they deliver the last mile. Currently, cable companies are building coax to residential locations, and traditional ISPs and phone companies are building copper to business locations. There is not a lot of activity thus far in building coax to business locations, so I would assume the telcos and traditional ISPs will dominate the businesses with T1/T3/xDSL service without much of a challenge from cable purveyors. I would also assume that xDSL will only threaten cable-modem purveyors for a short period until the cable operators get more of their pricing and physical-plant issues resolved. -rich - Subcription/unsubscription/info requests: send e-mail with "subscribe", "unsubscribe", or "info" on the first line of the message body to discuss-request at blu.org (Subject line is ignored).
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |