Re: Host Addressing Conventions [addr-conv 2.8]

Garrett.Wollman@uvm.edu Mon, 14 June 1993 21:48 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa11805; 14 Jun 93 17:48 EDT
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa11801; 14 Jun 93 17:48 EDT
Received: from thumper.bellcore.com by CNRI.Reston.VA.US id aa21929; 14 Jun 93 17:48 EDT
Received: by thumper.bellcore.com (4.1/4.7) id <AA09083> for ietf-archive@nri.reston.va.us; Mon, 14 Jun 93 17:47:44 EDT
Received: from sadye.emba.uvm.edu by thumper.bellcore.com (4.1/4.7) id <AA09079> for /usr/lib/sendmail -oi -fowner-pip X-pip; Mon, 14 Jun 93 17:47:40 EDT
Received: by sadye.emba.uvm.edu id AA17757 (5.65/6.02); Mon, 14 Jun 93 17:47:38 -0400
Date: Mon, 14 Jun 1993 17:47:38 -0400
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: Garrett.Wollman@uvm.edu
Message-Id: <9306142147.AA17757@sadye.emba.uvm.edu>
To: Paul Francis--formerly Tsuchiya <francis@thumper.bellcore.com>
Cc: pip@thumper.bellcore.com
Subject: Re: Host Addressing Conventions [addr-conv 2.8]
In-Reply-To: <9306142124.AA13220@tsuchiya.bellcore.com>
References: <9306142124.AA13220@tsuchiya.bellcore.com>

<<On Mon, 14 Jun 93 17:24:58 EDT, francis@thumper.bellcore.com (Paul Francis--formerly Tsuchiya) said:

> No, I mean LAN.  I really like LAN based addressing.  It
> seems to me a very natural place to put an address boundary--
> address follows topology.....

Well, from the point of view of the network I'm sitting on, it's no
more ``natural'' than considering our twelve different subnets all
part of one ``area'' (they are all connected to the same router) and
making that the bottom level of our addressing hierarchy.  This also
solves a bit of an administrative hassle that we have in terms of
files that have to get updated (remember, we're still talking
/near-term/ here) and so on, just to move one AUI cable from one jack
to another a foot away.

From the logical point of view, this set of twelve LANs functions as a
single unit, and I would dearly like to be able to address hosts on it
as if it were.

> As for multi-homing, it is solved in another way....the host gets
> multiple addresses, a Pip host picks one, and switches to the other
> if it gets a PCMP Packet Not Delivered indicating that one LAN
> can't be reached.

I would much rather have the router make this decision, based on---for
example---network utilization.  I think I sent an example of this
about two months ago, but I forget where.  The host really doesn't
have the proper information to make this choice.  As I noted in my
earlier message, I think that simple textual substitution of ``area''
for ``LAN'' will allow this to work, when combined with either some
form of Proxy ARP (ugh!) or an address-resolution mechanism that
looks like ESIS.

> (Anyway, even with areas, you haven't completely solved the
> multi-homing problem.  A host might be multi-homed to multiple
> areas).

Well, I would rather say that my underlying assumption is that, if a
host is in more than one area, then another host (at least at that
particular site) can make a meaningful policy decision between the two
areas for purposes of address selection.  The same is not true of
hosts which are connected to more than one wire but in the same area.

-GAWollman

--
Garrett A. Wollman   | Shashish is simple, it's discreet, it's brief. ... 
wollman@emba.uvm.edu | Shashish is the bonding of hearts in spite of distance.
uvm-gen!wollman      | It is a bond more powerful than absence.  We like people
UVM disagrees.       | who like Shashish.  - Claude McKenzie + Florent Vollant