Requirements for IP-in-IP tunnel MTU assurance

["Templin, Fred L" <Fred.L.Templin@boeing.com>]

Fred,

At the v6ops wg meeting during IETF64, I presented "Requirements
for IP-in-IP tunnel MTU assurance":

http://www.ietf.org/internet-drafts/draft-templin-mtuassurance-01.txt

Paraphrasing from the minutes (see below), the plan was for
the document to try to find a home in a different working
group such as pmtud or softwires then brought back to v6ops
if the other avenues didn't pan out. Since IETF64, there has
been some discussion in the other groups but apparently no
sustainable traction, so I am returning again to v6ops as
suggested. Please advise on how to proceed from here.

Thanks - Fred
fred.l.templin@boeing.com

Exceprt from IETF64 v6ops minutes:
**********************************
IP in IP tunnel MTU assurance (Fred Templin) 

	Draft presentation 

		Tunnel mechanisms have no means of assuring tunnel MTU 

		Issue: IPv4 fragmentation is bad, IPv4 path MTU is not
reliable (NAT, Firewall) 

		List of requirements 


	Dave Thaler: assurance might be too strong a word, we have no
assurance 
	on Ethernet for example, this is best effort. 

	Fred: best effort is what is meant, need to define the
terminology 

	Dave Thaler: this should be reviewed by the pmtud wg 

	Pekka: see some potential problems, maybe not as bad as author
sees. 

	Would be interested in working on this if this was done in the
venue that 
	will work on the solution 

	Fred: what would be the correct venue? Pmtud? 

	Pekka: maybe, need to look again at their charter 

	Dave Thaler: pmtud is the right place with the right expertise, 
	may require a charter update. Documenting the issue in v6ops 
	(and separating the requirement) is the right thing to do 

	Alain: the purpose of v6ops is to document operational issues, 
	so this document belongs here 

	Thomas Narten: Softwire may be the right place to fix the
problem 
	pmtud is focusing on TCP, this is a generic tunneling problem, 
	so softwire might be a better place. 

	Fred: go talk to chairs of pmtud and chairs of softwire if they
want to take it. 
	If this remains an issue, go back to v6ops and we will see with
AD what to do