On Nov 11, 2009, at 5:06 PM, Tony Li wrote:
Scott,
But it's not routing's problem that transport misuses location-
dependent
information. Lixia's quote is quite true.
The fact that transport is overloaded on top of routing makes it
routing's responsibility to get out from underneath. Transport is
fat dumb and happy where it is. If routing does not force the
change, then we have the (unacceptable) status quo.
Regards,
Tony
the fact that TCP uses IP address for its own connection identifier
makes it broke when the incoming interface address changes (either
host moved, or packets just coming from a different interface)
So all we need to do to fix this specific problem is to find TCP a
topologically independent identifier to use (e.g. HIP as one example,
though I do not mean to endorse HIP here, just as an example)
Lixia
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
