Re: [Tsvwg] WGLC for Port Randomization starts now (April 1st)

[Mark Allman <[e-mail]>]

My observation was merely meant to be quite high level.  I.e., an
end-point can prevent collisions by keeping state.  In TCP, half of this
is well-covered by the usual TIME-WAIT state.  Half of it would have to
be some new, local state a host decided to keep so that it avoided these
collisions.  All I was suggesting was that the port obfuscation I-D
could briefly note that keeping state is one way to avoid collisions if
the collision rate of the chosen port obfuscation technique is thought
to be problematic in some situation.  My intention was not to suggest
some big change here.  A sentence or two is what I was thinking.

(And, this should not be construed as support for
draft-ananth-tsvwg-timewait.) 

allman