[e2e] TCP, retransmissions, and ISPs with byte caps

[RJ Atkinson]

On Tuesday, June 18, 2002, at 12:35 , Philippe Strauss wrote:
> so, it's a technology independant problem: you have more
> probability of data corruption in the upstream, and the allocation
> scheme is more difficult to do well in the 500 -> 1 direction.

Phillipe,

	In a LANcity system, the upstream and downstream noise issues
are identical -- because upstream RF and downstream RF are tightly
coupled into a loop.  So the probability of data corruption is
about the same either way in that technology.  As I said earlier,
this is in part dependent on which specific CM/CMTS technology
one uses.

Ran