Re: [v6ops] new draft: draft-v6ops-pmtud-ecmp-problem

[Ray Hunter <[e-mail]>]

Fred Baker (fred) wrote:
> On Aug 25, 2014, at 4:47 AM, [e-mail] wrote:
>
>> A new draft has been posted, at http://datatracker.ietf.org/doc/draft-v6ops-pmtud-ecmp-problem/. Please take a look at it and comment.
>
> As noted, there’s a draft. We have, for the past week or so, had a discussion in the thread “PMTUD problem”.
>
> Question for the masses: what do we want to do with this draft? At IETF 90, we discussed adopting an updated version as a working group draft. Do we want to do that?
>
> More importantly, what outcome do we want. Does it need to become an RFC as operational viewpoint or procedure? Do we want to make recommendations to tcpm, 6man, or someone else, and specifically what recommendations?
I've read the draft.

I think the more generic problem statement is "Using ICMPv6 PTB 
signaling for path MTU discovery is currently unreliable" together with 
the related problem "fragmentation doesn't work even if the IPv6 stack 
knows the path MTU "

Whether that be due to:
-bad implementations
-ECMP
-filters
-tunnels
.....

The result is a global clamp of 1280 octets. And individual applications 
that have to perform their own application layer message fragmentation.

Would a draft summarizing the ICMP PTB problem, investigating potential 
alternative sources of path MTU information, and alternative signaling 
mechanisms for path MTU, fit within v6ops remit?

The (lack of) support for fragmentation already seems to be well known.

-- 
Regards,
RayH