Backwards compatability with existing IPv6 [was: Re: Network layer reqt? [was Re: Transport level multihoming]]

[Thomas Narten <narten@raleigh.ibm.com>]

> Considering that there virtualy is NO install base of IPv6, we
> can assume anything.

This is a point we'll discuss for a long time I'm sure, but I believe
it is impractical to assume there is no installed base and thus we are
free to do anything. The reality is that there are IPv6 products (for
end hosts) in the pipeline (Sun Solaris-8, Windows XP, AIX, etc.)
Getting *any* changes into end hosts will not happen until a minimum
of 2-3 years from today (i.e., any changes need to be made into RFCs,
and we don't have any proposals yet) and then would need to line up
with individual vendor product plans. Note that I am not necessarily
talking about *deployment*; I'm talking about products that include
IPv6 support. When the time comes that people turn them on, they will
simply turn on the version they already have. Thus, the boxes that
ship in the next few years will have an impact.

IMO, saying "no changes are out of scope" or "no changes can be
considered" misses the point. If/when there is a concrete proposal,
we'll need to weigh the advantages vs. the disadvantages.
Compatability issues with existing implementations will be one of the
factors that will need to be weighed.

Thomas