Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!ucsd!usc!samsung!rex!ukma!david
From: david@ms.uky.edu (David Herron -- One of the vertebrae)
Newsgroups: comp.protocols.tcp-ip
Subject: Re: ARP+
Keywords: ARP RIP RARP host
Message-ID: <13231@s.ms.uky.edu>
Date: 14 Nov 89 04:33:48 GMT
References: <1989Nov7.210118.13557@brutus.cs.uiuc.edu>
Reply-To: david@ms.uky.edu (David Herron -- One of the vertebrae)
Distribution: comp
Organization: U of Kentucky, Mathematical Sciences
Lines: 26

There was 3 or 4 RFC's in the 900's which were on the subject of
"transparent subnetting" ... one of the ideas there I thought had
a lot of promise but I don't think is implemented anywhere.  The idea
as I recall is to transmit your ARP as always.  The gateways take your
ARP and because it's subnet mask isn't right know it's not on the local
subnet and turn around and send it into the other networks they're
attached to.  The gateways on those subnets do the same look-at-subnet-
and-rebroadcast-if-necessary routine.

There was a couple of things in there which I don't recall too well
which kept loops from occuring.  Also I'm not 100% certain how the
response got back to the sender but it's probably by way of normal
IP routing.  Loop prevention could be done with a Usenet Path: style
marker in the packet which records the subnets each instance of the
packet has traversed, but that would break the protocol a bit.

This is tempting because it places all the complications in the
gateways and out of the hosts -- something that is likely a
desirable end goal.


-- 
<- David Herron; an MMDF guy                              <david@ms.uky.edu>
<- ska: David le casse\*'      {rutgers,uunet}!ukma!david, david@UKMA.BITNET
<- 
<- New official address:  attmail!sparsdev!dsh@attunix.att.com