From: Steve Schear (email@example.com)
Date: Fri Mar 10 2000 - 01:24:39 PST
At 03:37 PM 3/4/00 -0500, Steven M. Bellovin wrote:
>If you want to add more downstream bandwidth, it's easy -- just allocate
>another TV channel to the Internet instead of the Do-it-yourself Gerbil
>Network. But what do you do to add more bandwidth upstream? Given the
>of the two-way cable plants, your only choice is to split the fiber nodes.
I've been to dozens of cable plants and never have I seen one in which more
than a dozen MHz of the 40 MHz potential bandwidth is occupied. The
primary reasons, IMHO, is that few cable plants are properly designed for
broadband upstream use and few plant engineers and technicians are able to
adequately identify and correct field problems. The most significant
problem is ingress noise. Simple and economic RF solutions are available,
they are just not commonly being applied. I have only seen a two plants
where engineers really understood the theory and application of ingress
>That generally calls for many more fiber nodes, and hence the replacement
>lot of coax with fiber. Yes, it can be done, but it's expensive.
Fiber node splitting is one approach to increasing bandwidth, but reducing
the number of subscribers/node can help also increase overall system
reliability. It is one goal of AT&T mini-node architecture. Its also
aimed at solving the upstream path engineering/maintenance problems which
are not being addressed using conventional analog approaches.
>Bottom line -- there is certainly a business decision about how you allocate
>your limited upstream bandwidth. But it is much more limited than downstream
>bandwidth, and increasing it is much more expensive.
I have seen few plants with more than 2 or 3 (6MHz) unused downstream
channels. Most are loaded with the Home and Garden Channel and such. The
switch to digital has improved spectrum utilization but as always
programming increases to fill capacity. However, since data can (if
properly marketed) generate more revenue than all but the pay-per-view
channels all is not lost.
This archive was generated by hypermail 2b29 : Fri Mar 10 2000 - 09:25:03 PST