[e2e] A Question on the TCP handoff

Detlef Bosau detlef.bosau at web.de
Wed Nov 30 11:19:13 PST 2005

Alper Kamil Demir wrote:
> David,
> Thank you very much for the reply. However, I think tcpcp will not
> solve the problem because tcpcp is useful to migrate tcp connection
> from place to place (correct me if I am wrong). What I want to achieve
> is to replace the warm-up connection with an already
> established actual connection (so that replaced new connection
> both does have the same previous flow control of actual connection and
> doesn't go into slow start process by having congestion control state
> of warm-up).

Admittedly, I don´t quite understand what you mean by "warm up

However, IMHO there is some basic difficulty in any kind of TCP
handover, which even holds in the existing and well known approaches.

At least CWND is an estimator for a flow´s fair share of the
end--to--end _storage_ _capacity_ of a path.
Hence, each kind of "handover of a TCP connection from one path to
another", even approaches which attempt to recover from wrong /
congestion control issues by recovery of a former state (to my
understanding, this is done by TCP/Eifel) will only achieve its goal
when the recovered/trasfered
state variables are by some chance appropriate for the new path. I don´t
expect this to happen quite often in reality.

In my opinion, if the path of of a TCP flow changes, we cannot assume
the former CWND to be appropriate, however it´s easier to keep the old
value than to flip a coin for a new one. CWND is an estimator. So, if it
is correct, anything is fine. If not: It will converge to the correct
value due to TCP congestion control. Both cases are fine. So I don´t
really understand people who are eager to maintain old states for new

Detlef Bosau
Galileistrasse 30
70565 Stuttgart
Mail: detlef.bosau at web.de
Web: http://www.detlef-bosau.de
Mobile: +49 172 681 9937

More information about the end2end-interest mailing list