Document: draft-ietf-pwe3-satop-03.txt Reviewer: Elwyn Davies [elwynd@dial.pipex.com] Review Date: Monday 12/12/2005 4:48 AM CST Telechat Date: Thursday 12/15/2005 Summary: This is a re-review of an updated version of this document which I previously reviewed for IETF LC back in September 2005. Almost all of my previous concerns have been addressed by this version and it is almost ready for PS. Details: One remaining minor concern: The authors and I discussed the meaning of 'stray packets' in s6.3 since I was unclear what was meant by the term in this context. The dialogue was: ------------- > s6.3: I have no idea what a stray packet is or how I would > detect one! > The previous section talks about lost packets, so presumably > we are not talking about these, but I thought astray usually > might mean misrouted in which case the packet would never > arrive and can only be detected as lost. Does it mean > 'excessively misordered'? I am not sure what role the sync > source plays in this. A stray packet is a packet that would normally not reach the end-point, but did so anyway. For example, consider one SAToP PW being torn down and a new one over-expeditiously set up using the same PW label. Late packets from the old PW could be erroneously perceived as the packets of the new one, but could be detected when the optional RTP header is used by observing the SSRC value. ------- This explanation didn't (fully) make it into the update and I think it would help. This may also be tied into the comment on 'misconnection' in the Security Considerations (s9)? References: PWE3-ARCH, PWE3-IANA, PWE3-TDM-CONTROL and PWE3-VCCV are normative IMO as the protocol cannot be implemented without knowledge of these items. RFC1122 and RFC3551 aren't referenced. [RFC2112] s/b [RFC2212] Arguably RFC2212, RFC2914 and RFC3086 are informative, since they aren't needed to implement the protocol. Editorial: There are still a number of unexpanded acronyms in the document (MTU, L2TPv3, QoS, PDB leapt out at me). The terms PSN-bound and CE-bound are not defined. L2TPv3 could do with a reference. s4.3: s/smoothless/seamless/ maybe??? Smoothless is in interesting new word! s6.3, para 3rd from end: s/amount of times/amount of time/