Hello again,
>From the note below, it's obvious that there are some misconceptions
about Mobile IP that I'd like to correct.
Tripp Lilley wrote:
> That is, that's the way it should work... I should be able to put my
> laptop to sleep, hop on the plane, land in San Francisco, pull it back
> out, unsleep it, and pick up where I left off, _without_ tunneling all the
> way back home (ie: MobileIP). That's about as bad as going through MAE
> East just to get to the ISP down the street...
In the mobile-ip working group, we had many discussions in the years
before RFC 2002 about this problem. We also had a solution, called
"route optimization". The decision was taken to standardize the
"base protocol" and work on standardizing "route optimization"
later. The reason was exactly because doing this sort of route
optimization needs kernel modifications.
I think that all the details have been worked out pretty well for
Mobile IP route optimization. The specification has been implemented,
works, and may eventually get through last call if there is sufficient
interest. The strategy to isolate route optimization from the base
Mobile IP protocol may be questioned with the benefit of hindsight,
but after all there are now millions of implementations of Mobile IP
deployed (according to mail sent to the mobile-ip mailing list).
We don't know if that number would be bigger or smaller if the
working group had included route optimization features with
the base protocol.
I think it would be preferable to put effort behind a known pretty-good
approach that takes modifications to the networking stack, instead
of starting over on a relatively unknown approach that also requires
modifications to the networking stack. By known pretty-good, I mean
that it has a good story on security, has been studied for scalability,
and does the job. Many other approaches could be taken with more
engineering effort, but I don't think any of the other possibilties are
anywhere near as close to being finished. If there is something about
Mobile IP route optimization that should be improved, let's talk
about it; but let's also recognize that a lot of work has been done.
Regards,
Charlie P.
