RE: "Long Thin Networks Internet-Draft" submitted for PILC B

From: Spencer Dawkins (Spencer.Dawkins.sdawkins@nt.com)
Date: Mon Nov 23 1998 - 18:18:04 EST


Aaron,

this isn't a definitive delta, but it's close...

The biggest delta isn't actually that much of a change - we moved our
recommendations section into the previous section of the paper, proposal by
proposal. This matches more closely the approach taken by TCPSAT, which I
like.

This was the list of work items that Gab and I were using for coordination
(and if it doesn't say DONE, it's probably a next-time activity). In
addition, we also added a section on sharing TCP control blocks (4.13). And
we probably did a few other minor items.

        - in recommendations section (5) need to beef up:
                o Improved Scheduling Algorithms on End Systems
                  (was CBQ section)
                o spell out issues with delayed dupacks
                o why is ipsec a problem for ecn but not for
                  snoop? - DONE (although I think the section
                      for snoop was already explicit enough)

        - need references for cdpd and gsm/gprs - DONE

        - check for changes in "stability of the proposal"
          in the table, as some stuff may already be rfc
          (e.g., header compression/ payload compression..) DONE - but I
thought
               IPPCP was approved as PS, and it's still showing up as IDs on
the WG home page...
        
        - add more verbage on why van jacobson compression is bad
        for us, and why degermark et al have a better alternative,
        and recommend using it - DONE
 
        - add verbage mentioned by hari about ack-every-segment being
        bad for asymmetric networks, cuz it effectively limits the
        forward bandwidth - DONE
        
        - add pointer and ref to scps (mentioned by travis)
                http://www.scps.org/scps

        - fetch refs in badri's protocol adaptation paper
        - add section on M-TCP (not MTCP)
        - add section on Amigos
        - add verbage on disconnections and burst errors/handoffs
        - draft-ietf-tcpsat-*.txt - standard mechanisms DONE, research DONE
        - draft-kksjf-ecn-*.txt - DONE
        - draft-rfced-info-partridge-*.txt - DONE
        - rfc 2414 (was draft-floyd-incr-init-win-*.txt) - DONE
        - rfc 2415 (was draft-ietf-tcpimpl-poduri-*.txt) - DONE,
        - rfc 2416 (was draft-shepard-tcp-4-packets-3-buff-*.txt) - DONE
        - draft-ietf-tcpimpl-prob-*.txt - DONE but not included
        - draft-ietf-tcpimpl-cong-control-*.txt
                replaces rfc2000, stevens - DONE
        - modify considerations (e.g. on snoop) based on nitin's
                simulations



This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 09:12:19 EST