Re: ipv6nd-02.txt & autoconfig

schulter@zk3.dec.com Thu, 09 May 1996 17:41 UTC

Received: from ietf.cnri.reston.va.us by IETF.CNRI.Reston.VA.US id aa19693; 9 May 96 13:41 EDT
Received: from CNRI.Reston.VA.US by IETF.CNRI.Reston.VA.US id aa19689; 9 May 96 13:41 EDT
Received: from guelah.nexen.com by CNRI.Reston.VA.US id aa11958; 9 May 96 13:41 EDT
Received: from maelstrom.nexen.com (maelstrom.nexen.com [204.249.97.5]) by guelah.nexen.com (8.7.3/8.7.3) with ESMTP id NAA16917; Thu, 9 May 1996 13:33:23 -0400 (EDT)
Received: (from root@localhost) by maelstrom.nexen.com (8.7.3/8.7.3) id NAA11235 for ip-atm-out; Thu, 9 May 1996 13:24:11 -0400 (EDT)
Received: from guelah.nexen.com (guelah.nexen.com [204.249.96.19]) by maelstrom.nexen.com (8.7.3/8.7.3) with ESMTP id NAA11224 for <ip-atm@nexen.com>; Thu, 9 May 1996 13:24:07 -0400 (EDT)
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: schulter@zk3.dec.com
Received: from mail13.digital.com (mail13.digital.com [192.208.46.30]) by guelah.nexen.com (8.7.3/8.7.3) with ESMTP id NAA16787 for <ip-atm@nexen.com>; Thu, 9 May 1996 13:24:05 -0400 (EDT)
Received: from hunch.zk3.dec.com by mail13.digital.com (8.7.5/UNX 1.2/1.0/WV) id NAA12156; Thu, 9 May 1996 13:00:23 -0400 (EDT)
Received: from dogfish.zk3.dec.com by hunch.zk3.dec.com; (5.65v3.2/1.1.8.2/11Mar96-0342PM) id AA06304; Thu, 9 May 1996 12:59:52 -0400
Received: from localhost by dogfish.zk3.dec.com (5.65v3.2/1.1.10.3/27Jun95-1215PM) id AA07972; Thu, 9 May 1996 12:59:48 -0400
Message-Id: <9605091659.AA07972@dogfish.zk3.dec.com>
X-Mailer: exmh version 1.6.1 5/23/95
To: Albert Manfredi <manfredi@engr05.comsys.rockwell.com>
Cc: gja@bellcore.com, IP-ATM Working Group <ip-atm@nexen.com>
Subject: Re: ipv6nd-02.txt & autoconfig
In-Reply-To: Your message of "Thu, 09 May 96 10:31:40 PDT." <31922BFC.7019@engr05.comsys.rockwell.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Thu, 09 May 1996 12:59:48 -0400
X-Mts: smtp
X-Orig-Sender: owner-ip-atm@nexen.com
Precedence: bulk
X-Info: Submissions to ip-atm@nexen.com
X-Info: [Un]Subscribe requests to majordomo@nexen.com
X-Info: Archives via http://cell-relay.indiana.edu/cell-relay/archives/IPATM/IPATM.html

on Thu, 09 May 96 10:31:40 PDT Albert Manfredi wrote:

> I guess my point is that the SEL should be viewed as an ATM-only neat-o
> feature. It should not be used to map to different IP addresses.

Well, why not, especially if we want to come up with a solution that takes
optimal advantage of ATM (and not a generic NBMA solution).  This is
a neat feature that ATM has which we can exploit to provide different
virtual IP subnets using the least number of network addresses.  BTW, my view
is that we should be trying to do the best thing for ATM, and not
limiting our solutions by other, less capable, NBMA technologies.

My view on the selector byte is that the selector identifies a specific
service (or instance of a service) on the end node.  

Perhaps we're also confusing two issues here.  That is, virtual subnets
and virtual hosts.  In a case where a host wants to join multiple links
it is sufficient to use one ESI and use the selector to identify the
subnet on the host.  In this case, whatever host token is generated could
be the used on each subnet, so the ESI _might_ be sufficient as long as
it is unique on each subnet.  In the case of virtual hosts where one
end node is actually masquerading as different hosts, then this is
a case where we would probably want a different ESI for each virtual host,
but then each virtual host could join multiple virtual links and use
the SEL byte to differentiate the virtual links.

I think a restriction of one logical link per ATM interface is too restrictive
and too expensive.  There is no reason that a node that wants to join
different links should not be able to do this through a single adapter.
To do otherwise would not be taking full advantage of ATM capabilities, IMHO.

> A new host boots up. This guy cares what LLG it joins. It listens on
> VCI=x until the server transmitting the proper IPv6 (and ATM?) routable
> prefix sends its autoconfig message.

What is a proper IPv6 routable prefix?  How would an IPv6 host know which
LLG (or Logical Link ;-) it wants to join ahead of time based only on
IPv6 address information?  Prefixes on links can change.  Links need not
have any prefixes.  A node that uses autoconf or DHCPv6 won't have any
idea what its IPv6 prefixes are at the time it tries to connect to a 
link.

I think what we need, especially for IPv6, is some sort of LLG (or LL)
cookie which uniquely identifies a link independent of any network
layer address information.  This is something NBMA or ATM specific which
isn't required by other interconnects, and which should not be visible
to IPv6 (it's part of the IPv6 to ATM convergence configuration).

What about using ILMI or a configuration server to find servers for specific
services?  I really think using a well known VC would have too many scalability
problems.

 --- pete

------------------
Peter Schulter					schulter@zk3.dec.com
Digital UNIX Networking				voice (603) 881-2920
Digital Equipment Corp				voice (DTN) 381-2920
ZK3-03/U14					FAX   (603) 881-2257
110 Spit Brook Road				FAX   (DTN) 381-2257
Nashua, NH 03062