>Delayed dupacks is in the second class -- the schemes in the
>second class are "end-to-end" since they cannot get much help
>from the network. As such, one should expect these schemes
This is an example of a proposed change to TCP that does make sense,
because it deals with a generic (if not yet common) Internet problem
(out of order packet delivery) rather than something highly specific
to one particular class of (radio) links.
The fast recovery mechanism in TCP has always bothered me a little
precisely because its reaction to out-of-order delivery. The Internet
architecture has always permitted this, even if it doesn't happen all
that often in practice (yet). Even without radio links, the increased
use of load splitting will make this a problem worth solving at the
TCP level.
You could also avoid this problem with a link ARQ protocol that
enforces in-order delivery, at least for packets that use the ARQ
mechanism (e.g., TCP segments that set the TOS Reliability bit).
Phil
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 09:12:20 EST