Re: Routing IPv4 with OSPFv3

Yasuhiro Ohara <yasu@SFC.WIDE.AD.JP> Thu, 15 August 2002 08:17 UTC

Received: from cherry.ease.lsoft.com (cherry.ease.lsoft.com [209.119.0.109]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id EAA16879 for <ospf-archive@LISTS.IETF.ORG>; Thu, 15 Aug 2002 04:17:10 -0400 (EDT)
Received: from walnut (209.119.0.61) by cherry.ease.lsoft.com (LSMTP for Digital Unix v1.1b) with SMTP id <6.006D5F14@cherry.ease.lsoft.com>; Thu, 15 Aug 2002 4:18:28 -0400
Received: from DISCUSS.MICROSOFT.COM by DISCUSS.MICROSOFT.COM (LISTSERV-TCP/IP release 1.8e) with spool id 109353 for OSPF@DISCUSS.MICROSOFT.COM; Thu, 15 Aug 2002 04:18:24 -0400
Received: from 203.178.142.130 by WALNUT.EASE.LSOFT.COM (SMTPL release 1.0f) with TCP; Thu, 15 Aug 2002 04:18:24 -0400
Received: from localhost (page.sfc.wide.ad.jp [203.178.143.89]) by shonan.sfc.wide.ad.jp (Postfix) with ESMTP id DB27D5D019; Thu, 15 Aug 2002 17:18:25 +0900 (JST)
References: <3D5B4B88.215EA127@arc.corp.mot.com>
X-Mailer: Mew version 2.1 on XEmacs 21.1.14 (Cuyahoga Valley)
Mime-Version: 1.0
Content-Type: Text/Plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Message-ID: <20020815.171008.123934010.yasu@sfc.wide.ad.jp>
Date: Thu, 15 Aug 2002 17:10:08 +0900
Reply-To: Mailing List <OSPF@DISCUSS.MICROSOFT.COM>
Sender: Mailing List <OSPF@DISCUSS.MICROSOFT.COM>
From: Yasuhiro Ohara <yasu@SFC.WIDE.AD.JP>
Subject: Re: Routing IPv4 with OSPFv3
Comments: To: arthurd@ARC.CORP.MOT.COM
To: OSPF@DISCUSS.MICROSOFT.COM
In-Reply-To: <3D5B4B88.215EA127@arc.corp.mot.com>
Precedence: list
Content-Transfer-Encoding: 7bit

Hi.

arthurd> Greetings,
arthurd>
arthurd> I'm interested in people's opinions and experiences in using
arthurd> OSPFv3 (aka OSPF for IPv6) for routing IPv4.
arthurd>
arthurd> My understanding from reading the OSPFv3 specification
arthurd> (RFC2740) is that OSPFv3 is designed with a degree of
arthurd> protocol independence. By this I mean that you could, in
arthurd> principle, use OSPFv3 to route any protocol family you chose,
arthurd> just so long as you were able to map addressing information
arthurd> to the topology (and of course your OSPF code knew how to set
arthurd> up the correct forwarding state for your given protocol
arthurd> family). The topology-to-address mapping for IPv6 is done
arthurd> using OSPFv3's Intra-area-prefix-LSAs, but the spec makes no
arthurd> mention of how you might map IPv4 addressing information to a
arthurd> network topology.

If you care only about IPv4, it is possible without involving protocol
issue, since there are IPv4-mapped address in IPv6 address
architecture (specification). You can simply advertise IPv4 prefix in
IPv6 "IPv4-mapped" format, then the receiver of the advertisement can
install the prefix in IPv4 route table.

I agree that OSPFv3 can be adopted to any Network Address Family, by
defining new LSAs corresponds to Prefix-LSAs including Link-LSA.

arthurd> So, my questions are:
arthurd> - Regarding OSPFv3 - is the bulk of it protocol agnostic, or
arthurd> is it just my imagination? Was it the intention of the
arthurd> protocol authors to create a protocol that could easily route
arthurd> multiple protocol families, or just IPv6?

I guess it was the authors intention (I read John Moy's interview
somewhere mentioning it). But there's no "protocol family specifier"
field in the LSA, so we must indicate protocol family (address family)
implicitly by LSA type.

arthurd> - Is there any interest out there in using OSPFv3 to route
arthurd> IPv4?

I can't understand the merit of IPv4 routing by OSPFv3. I admit that
we can easily route IPv4 by OSPFv3.

arthurd> - Do people think it is a good idea in general to have a
arthurd> single protocol route both v4 and v6? (I'm guessing that
arthurd> there are those out there who think that OSPFv2 works just
arthurd> fine so why would you want to change it ... )

This will lead to "ships-in-the-night" discussion. IMHO I think it is
better to caluculate route individually in each address family. I am
happy with two independent routing running, as I can debug IPv6
routing using IPv4 remote login, and even also I can debug IPv4
routing using IPv6 remote login !
[* ships-in-the-night is mentioned in 6.2.5 of the book "Routing in
the Internet"]

arthurd> And while I'm at it, are there any such things as Opaque LSAs for
arthurd> OSPFv3?

It is built-in feature of OSPFv3.

regards.
yasu