Re: Another proposal to think about

smb@hector.att.com Thu, 23 November 1989 15:03 UTC

Received: from decwrl.dec.com by acetes.pa.dec.com (5.54.5/4.7.34) id AA09530; Thu, 23 Nov 89 07:03:31 PST
Received: by decwrl.dec.com; id AA11038; Thu, 23 Nov 89 07:03:01 -0800
From: smb@hector.att.com
Message-Id: <8911231457.AA22278@hector.homer.nj.att.com>
Received: by hector.homer.nj.att.com id AA22278; Thu, 23 Nov 89 09:57:44 EST
To: MTU Discovery <mtudwg>
Subject: Re: Another proposal to think about
Date: Thu, 23 Nov 89 09:57:43 EST
>From: hector!smb

If there are objections to using a bit in the IP header for the ``Report
Fragmentation'' flag, what about using an IP option?  Unlike the
Mogul/Kent/Partridge/McCloghrie option, this would be end-to-end; only
the receiving host would act on it.  The semantics would be the same
as the Deering bit.  It's a bit harder to implement, since you do not
want to put an option into every packet, and since you'd have to account
for the option length in calculating the true path MTU (and in probing).


		--Steve Bellovin
		smb@ulysses.att.com