Re: [re-ECN] ConEx and MPLS (WAS RE: re-ECN Digest, Vol 14, Issue 18)

[ken carlberg <[e-mail]>]

Hi Ingemar,

On Apr 29, 2010, at 4:34 AM, Ingemar Johansson S wrote:

> I am too a bit curious about this as I belive that this is probably an important matter. 
> MPLS was (briefly) discussed on the list earlier
> http://www.ietf.org/mail-archive/web/re-ecn/current/msg00580.html
> As I understand it MPLS is out of scope because there is some pressure to prove that it is useful before attempts are made to adapt other protocols (like MPLS) to work with ConEx.

this last part is what concerns me about how the issue is brought up.  While I agree with the point that Stewart brings up in that MPLS is an integral part in a significant number of ISPs, a proper response to Stewart's question really requires a solution first.  

Stewart, I'm going by memory here, so please  correct me if I'm wrong, but the crux of re-ecn is the use of the evil bit as a means of distinguishing it from ECN.  So, wouldn't it be the case that transits that don't want to add any additional attribute to the LSP would just choose to disregard IP packets that have that bit set -- at least with respect to IPv4.  So, while it would eventually be constructive to address the MPLS case, the group and transits can by-pass that topic for the time being.

And if I recall correctly, it was stated and agreed at the Anaheim meeting that Conex needs to make sure its doesn't adversely affect MPLS.  At the very least, the IESG can look at the drafts and make the determination then if Conex adversely affects MPLS....but this is done after work/solutions  have been worked on instead of design-on-the-fly.

-ken