Re: I-D Action: draft-ietf-tcpm-1323bis-02.txt

Hi,

I have been asked by the chairs to act as editor on this draft, to shepard it and help finalize it.

The main differences between the last version published in March 2009 are summarized here:

Moved from nroff to xml2rfc.
  This changed the layout (indentation, TOC, pagination) somewhat.

The content changed in these three aspects, after hallway discussions at the last IETF and email discussions:

   RTTM update processing explicitly excludes packets containing
        SACK options.  This addresses inflation of the RTT during
        episodes of packet loss in both directions.

(This really addresses an omission in all known implementations on the spirit of RTTM).


    In Section 3.2 the if-clause allowing sending of timestamps only
        when received in a <SYN> or <SYN,ACK> was removed, to allow for
        late timestamp negotiation.

(to continue the ongoing WG discussion on TCP Option space exhaustion and possible (if temporary) mitigations)


 Section 2.4 was added describing the unavoidable window
        retraction issue, and explicitly describing the mitigation steps
        necessary.

(-01 already had a pointer to RFC1122 in here, but it may be helpful to implementers to have the proper action described once more explicitly in this document too).

Note: I submitted a version without the Nits-fix from Alexander (http://www.ietf.org/mail-archive/web/tcpm/current/msg04477.html) for some reason; this is fixed. Also, the issue brought up here may need more discussion, especially if Timestamps should become useful for more than just PAWS and coarse RTTM:

http://www.ietf.org/mail-archive/web/tcpm/current/msg04706.html



Best regards,
   Richard


-----Original Message-----
From: Scharf, Michael (Michael)
Sent: Mittwoch, 09. Mai 2012 14:10
To: tcpm@...

Dear all,

there have been suggestions to add an additional author to draft-ietf-tcpm-1323bis-01 in past TCPM meetings.

As a consequence, the chairs suggest to add Richard as an additional author in order the help with the completion of this WG item. Please let us know if this should result in any issue.

The understanding of the chairs is that the 1323bis document should really focus on fixing corner cases in RFC 1323, and we think that this document should be finished as soon as possible.

The recent discussion has shown that TCP timestamps could be further improved, e. g., by late activiation of options. However, this is a new concept that seems to require further studies and experiments. Our recommendation is therefore to address such new, interesting use of timestamps in a separate document.

Thanks

Michael (TCPM co-chair)
_______________________________________________
tcpm mailing list
tcpm@...
https://www.ietf.org/mailman/listinfo/tcpm