[secdir] brief comment on draft-ietf-pce-pcep-p2mp-extensions-09

[Sandra Murphy <[e-mail]>]

This draft caught my eye and I would have loved to investigate it fully. 
However, this is one of those drafts that requires a whole lot of advanced 
knowledge, and lacking that a whole lot of background reading.  I was not 
able to do that.

This is an extension ot the PCEP to provide for point-to-multipoint LSPs. 
Yep, multicast.  PCEP is the Path Comuptation Element Protocol (the PCE 
computes LSPs on behalf of routers in an AS or inter-ASs).  OSPF and IS-IS 
have been enhanced to help in the PCE discovery process (also in carrying 
around info for traffic engineered LSPs.).

A complicated set of background: MPLS, PCEP, P2MP, OSPF, ISIS, RSVP, 
yada yada.

(Excuses, excuses.)

But I did notice one thing I wanted to comment on. PCEP requires use of 
TCP-MD5 with recognition of its failings, specifically mentioning that 
TCP-AO was not yet complete.

This document refers to the PCEP security analysis.  But as TCP-AO has now 
passed the IESG, should there be some recognition of that change?

This doc does not seem the proper place to change the PCEP recommendation, 
but a change in the MUST requirement in PCEP (RFC5440) seems to be 
something to consider somehow.

I expect that's a AD level discussion between the security ADs and the 
routing ADs.

--Sandy