minutes of last week's meeting

mogul (Jeffrey Mogul) Tue, 08 May 1990 20:08 UTC

Received: by acetes.pa.dec.com (5.54.5/4.7.34) id AA09934; Tue, 8 May 90 13:08:48 PDT
From: mogul (Jeffrey Mogul)
Message-Id: <9005082008.AA09934@acetes.pa.dec.com>
Date: 8 May 1990 1308-PDT (Tuesday)
To: mtudwg
Subject: minutes of last week's meeting

Those of you who missed the meeting: this is what happened.

Those of you who were there: tell me ASAP if this is not what happened.

	MTU Discovery Working Group
	Chairperson: Jeffrey Mogul/DECWRL

	2 May 1990
	Held at the IETF meeting in Pittsburgh, PA
	Reported by Jeffrey Mogul


		a) Report on current draft (Mogul/Deering)
		b) Obtain consensus on approach
		c) Focus on details
		d) Schedule for standardization


	Fred Bohle		fab@saturn.acc.com
	Steve Brunigos(?)
	David Burdelski		daveb@ftp.com
	Duane Butler		dmb@network.com
	Andrew Cherenson	arc@sgi.com
	Noel Chiappa		jnc@PTT.LCS.MIT.EDU
	Steve Deering		deering@pescadero.stanford.edu
	Dave Forster
	Rich Fox		sytek!rfox@sun.com
	Karen Frisa		karen@kinetics.com
	Steve Hubert		hubert@cac.washington.edu
	Van Jacobson		van@helios.ee.lbl.gov
	Stev Knowles		stev@ftp.com
	Yoni Malachi		yoni@cs.stanford.edu
	Keith McCloghrie	sytek!kzm@hplabs.HP.COM
	Leo J. McLauglin III	ljm@twg.com
	Jeff Mogul		mogul@decwrl.dec.com
	Drew Perkins		ddp@andrew.cmu.edu
	John Moy		jmoy@proteon.com
	Stephanie Price		cmcvax!price@hub.ucsb.edu
	Mike Reilly		reilly@nsl.dec.com
	Tim Seaver		tas@mcnc.org
	Frank Slaughter		fgs@shiva.com
	Richard Smith		smiddy@pluto.dss.com
	Brad Strand		bstrand@cray.com
	Cal Thixton		cthixton@next.com
	John Veizades		veizades@apple.com
	Jonathan Wenseur(?)	jhw@shiva.com

	[Some of these people did not sign the roster, but were obviously
	there.  Some other people who didn't sign the roster have slipped
	my memory.]


	This was the third meeting of the MTU Discovery Working Group.

	Jeff Mogul started with a review of where we had been in the past 5
	months, including all the failed approaches.  He then presented
	the current proposal, originated by Steve Deering.  
	Summary: send all packets with the DF bit set.  Routers that
	cannot forward these packets return a slightly modified ICMP
	message that indicates the appropriate MTU.  (Current routers
	return "0" in the field meant for this purpose.)  The sending
	host revises its estimate of the Path MTU, and retransmits
	the dropped datagram.
	There was no objection to the basic design.  Some discussion
	ensued concerning details of the implementation and the
	relation between Path MTU discovery and transport protocol
	This design solves many problems with the previous designs,
	especially because it is compatible with existing hosts
	and routers, and is quite simple to implement.

	(1) Jeff Mogul and Steve Deering will change some details
	in the current draft and submit it as an Internet Draft.

	(2) The Router Requirements working group should be notified
	that we no longer care how fragmentation is done, since we
	don't rely on the sizes of fragments.  We will also recommend
	that support for MTU Discovery be made a requirement.

	We expect never to meet again.

	Progress towards standardization will go as fast as possible,
	unless serious objections are raised.