[Jim Bound]

Shiva,

 >   Thanks a lot for the reply.  I am aware of the fact that automatic
>tunnelling is not very prevalent in current designs. But I am trying to
>concentrate on "The Flag Day" , when we go in for a global transition to
>IPv6. But I have not yet been able to arrive upon a single problem statement
>on which to work. I do have a couple of more questions.

I don't think a flag day will ever work no matter how minimal the nodes
are for the transition except for a home user and the ISP is now doing
IPv6 and even then assuming all the apps have been ported to IPv6 would
have to be an a priori.  I believe all transition mechanisms defined in
the IETF and out of the IETF all assume no flag day.  Even if lets say
cell devices moved to IPv6 (which seems logical to many of us) they
would still have to interoperate with IPv4 for some parts assuming all
infrastructure was able to utilize Ipv6 for Wireless and Mobile paradigms 
in a specific geo-market segment (e.g. Sprint, Deutch Telecom, Singapore
Telecom, etc).

>    I was wondering as to how the priority field of a IPv6 packet gets
>translated while being encapsulated into an IPv4 packet, so that the same
>QoS is maintained even within the tunnel. Also how do we ensure that all the
>IP packets of the same flow label follow the same path and pass through the
>same tunnel?

If you mean the traffic class field diff serve group has done a good job
keeping them the same.  So that IP header field should mean the same
thing for most cases.  Though I have to read the latest specs on
aggregation of diff serve with RSVP.

If you mean the flow label this is still not officially defined except
for RSVP via the Flow Spec.  And this will only work with IPv6 capable
nodes.  If we do define a means to pass thru as you say the Ipv6
flowlabel plus the dst address, it will not be possible today if the
packet is encapsulated within IPv4 packet to maintain the same state
defined by the address+flowlabel in IPv6.  I say today because I guess
it would be possible to define some IPv4 "option" to carry the IPv6
flowlabel in the Ipv4 packet or possible in a route proto-id TBD.  But
getting that kind of idea done in IPv4 I think has zero chance for
success in the market unless Ipv6 gets much more deployed and then maybe
something like that may happen as a defined assistance transition
mechanism.

But if you have ideas to help this along I think all of us are all ears.

/jim