CURRENT_MEETING_REPORT_
Reported by Philip Almquist/Consultant
RREQ Minutes
The Router Requirements Working Group held extensive meetings in Atlanta
in an attempt to resolve the remaining outstanding issues. We were
mostly very successful: although a considerable number of loose ends
remain, we expect that they can be handled on the mailing list (and, if
necessary, in a videoconference). Our goal is to have a final version
of the document available in October to be formally submitted to the
standards process at the November IETF meeting.
The meetings focused on three activities:
1. Review of the Router Requirements draft (Monday, Wednesday, and
Thursday).
2. Discussion of the theoretical basis of routing in border routers
(Tuesday and Wednesday).
3. Review of a draft document describing IP type of service (Tuesday).
Some visitors from the IPLPDN Working Group asked us to consider the
changes to the IP routing architecture that they are considering, but we
did not have time to hold the extensive discussion which would have been
necessary to reach consensus on that issue.
Each is described in more detail below. The Chair would like to thank
Frank Solensky for recording the decisions reached during the meeting.
Review of Router Requirements Draft
The entire document was reviewed in detail. This process was
considerably less contentious than at many previous meetings, since the
most divisive issues had previously been resolved. Some of the issues
included:
o The relationship between our document and the Host Requirements,
and the extent to which our document ought to replicate material
found in the Host Requirements.
o When (if ever) a router should believe ICMP Redirects.
o Metrics for static routes.
o Whether SNMP may be implemented via a proxy agent.
o The security of in-band configuration mechanisms.
o The allowability of IP multicast addresses in source routes.
o What still needs to be done to complete the document.
1
In regard to the first issue, we concluded that overlap is generally not
desirable, and that we should work to eliminate it. The resolution of
the other issues should be obvious from the next version of the draft.
Theory of Border Routers
The Working Group continued its discussion from previous meetings of how
a router which is in multiple routing domains can choose from among
routes to the same destination learned from different routing protocols
(``route believability'') and how can it pass routing information
between multiple routing domains (``route leaking''). Our discussion of
these difficult topics was guided by several papers by Working Group
members (all available as Internet Drafts): ``Ruminations on the Next
Hop'' and ``Ruminations on Route Leaking'', both by Philip Almquist, and
``Some Thoughts on Multi-Domain Routing'' by Ross Callon.
The Working Group concluded that the discussion in the papers was
generally correct. However, two issues were raised concerning the first
paper:
1. The meta-lookup algorithm always picks the most specific route to
any destination; administrative policy controls are used only to
choose among routes that are equally specific. Some felt that some
network managers may wish to have policy influence route choice
among routes that are not equally specific (in particular, some may
wish to emulate the result of the ``Rank Ordering of Routing
Protocols'' approach).
2. The preference of a route is not influenced by route leaking. Some
felt that this is incorrect: the preference of a route ought to
improve if the route is leaked into a routing protocol whose
default preference value is better than the original preference
value of the route.
Neither of these issues were completely resolved at the meeting, so the
author was tasked to consider them in the next version of the paper.
IP Type of Service
Philip Almquist is attempting to write an RFC on the use of the TOS bits
in the IP header. The group briefly reviewed the then-current draft
(``Type of Service in the Internet Protocol'', available as an Internet
Draft). Although the group raised some editorial concerns, consensus on
the technical content was reached with almost no debate.
Attendees
Philip Almquist almquist@jessica.stanford.edu
William Barns barns@gateway.mitre.org
Arthur Berggreen art@acc.com
David Bridgham dab@asylum.sf.ca.us
2
Gregory Bruell gob@shiva.com
Ross Callon callon@bigfut.enet.dec.com
Vinton Cerf vcerf@nri.reston.va.us
Peter Chang tpc@mtunm.att.com
Rob Coltun rcoltun@ni.umd.edu
Kurt Dobbins dobbins@ctron.com
Robert Elz kre@munnari.oz.au
Dino Farinacci dino@cisco.com
Dennis Ferguson dennis@canet.ca
William Fink bill@wizard.gsfc.nasa.gov
David Forster forster@marvin.dec.com
Vince Fuller vaf@stanford.edu
Deborah Futcher dfutche@eco.twg.com
Chris Gunner gunner@osicwg.enet.dec.com
Steven Hotz hotz@isi.edu
B.V. Jagadeesh bvj@3com.com
Frank Kastenholz kasten@europa.clearpoint.com
Michael Khalandovsky mlk@ftp.com
Paulina Knibbe knibbe@cisco.com
Stev Knowles stev@ftp.com
Richard Larkin rblarkin@sprintf.merit.edu
John Lekashman lekash@nas.nasa.gov
Mark Lewis mlewis@telebit.com
Tony Li tli@cisco.com
John Lynn lynn@ttcllcat.cit.cornell.edu
Shane MacPhillamy slm@netrix.com
Gary Malkin gmalkin@ftp.com
Bill Manning bmanning@rice.edu
Bernie May bfm@houxa.att.com
April Merrill abmerri@tycho.ncsc.mil
John Moy jmoy@proteon.com
Karen O'Donoghue kodonog@relay.nswc.navy.mil
David O'Leary oleary@sura.net
Radia Perlman perlman@radia.enet.dec.com
Jason Perreault perreaul@interlan.interlan.com
Lars Poulsen lars@cmc.com
James Rees rees@ifs.umich.edu
Michael Reilly reilly@nsl.dec.com
Gershon Schatzberg 439-3582@mcimail.com
John Scudder jgs@merit.edu
John Seligson johns@ultra.com
Richard Smith smiddy@pluto.dss.com
Frank Solensky solensky@clearpoint.com
Emil Sturniolo emil@dss.com
Osamu Takada takada@sdl.hitachi.co.jp
Ross Veach rrv@uiuc.edu
Jonathan Wenocur jhw@shiva.com
Walter Wimer walter.wimer@andrew.cmu.edu
Cathy Wittbrodt cjw@nersc.gov
John Wobus jmwobus@suvm.acs.syr.edu
Robert Woodburn woody@cseic.saic.com
Joseph Zur fibrontics!zur@uunet.uu.net
Peter de Vries peter@wco.ftp.com
3