Re: [Roll] RA-DIOs and next-door destinations

"Pascal Thubert (pthubert)" <pthubert@cisco.com> Tue, 04 August 2009 13:07 UTC

Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Date: Tue, 04 Aug 2009 15:07:49 +0200
Message-ID: <7892795E1A87F04CADFCCF41FADD00FC07E20A2C@xmb-ams-337.emea.cisco.com>
In-Reply-To: <B8949F83-E028-4CAE-B5E4-E74EC5531F66@archrock.com>
Thread-Topic: [Roll] RA-DIOs and next-door destinations
Thread-Index: AcoUceaG2akn2vZaTb+QPgnVp04jiAAkVLfg
References: <87prbc5231.fsf@kelsey-ws.hq.ember.com> <B8949F83-E028-4CAE-B5E4-E74EC5531F66@archrock.com>
From: "Pascal Thubert (pthubert)" <pthubert@cisco.com>
To: Jonathan Hui <jhui@archrock.com>, Richard Kelsey <richard.kelsey@ember.com>
Cc: roll@ietf.org
Subject: Re: [Roll] RA-DIOs and next-door destinations
Precedence: list

Hi Jonathan:

If a node is a router A that owns a prefix A::/64 is near to router B that owns B::/64.

With the current text in the draft, A will advertise A::/64 to its parent using ND-DAO, and B will do the same for B::/64.
If they are lucky and are siblings sharing a same parent, then A::1 (might be A's address on A::/64) could ping B::1 via that parent. But the DAG might provide a path that's a lot worse than that.

An easy addition to the draft is to allow A to also multicast its NA-DAO to all nodes. That way, if B is in line of sight from A, B could install a route to A::/64 via A's link local regardless of their DAG relationship and forward the packets directly.

This is an easy way to address local communication between nodes. Note that if A is a host, A would advertise a host route A::1/128. 

What do you think?

Pascal

>-----Original Message-----
>From: roll-bounces@ietf.org [mailto:roll-bounces@ietf.org] On Behalf Of Jonathan Hui
>Sent: lundi 3 août 2009 21:38
>To: Richard Kelsey
>Cc: roll@ietf.org
>Subject: Re: [Roll] RA-DIOs and next-door destinations
>
>
>Hi Richard,
>
>On Aug 3, 2009, at 8:50 AM, Richard Kelsey wrote:
>> In the section on RA-DIO reception, draft-dt-roll-rpl-01
>> says that a RA-DIO from a neighbor that is not a potential
>> parent or a sibling should be ignored.  Is it permissable
>> to remember the neighbor in order to route messages to
>> it directly?  In other words, can a packet whose
>> destination is a neighbor with a greater depth be sent
>> directly to that neighbor?
>
>
>Let me speak to your comment directly, rather than the draft.
>
>I don't see any problem using the RA-DIO to figure out what neighbors
>are around you. One issue is that if the RA-DIO source address is a
>link-local address, it may not be so useful if you're trying to talk
>to that node using its global address. What is more useful is to allow
>including some kind of DAO in the same RA so that you know what
>addresses (other than the source address) are assigned to that node's
>interface.
>
>--
>Jonathan Hui
>
>_______________________________________________
>Roll mailing list
>Roll@ietf.org
>https://www.ietf.org/mailman/listinfo/roll

[Roll] RA-DIOs and next-door destinations Richard Kelsey
Re: [Roll] RA-DIOs and next-door destinations Jonathan Hui
Re: [Roll] RA-DIOs and next-door destinations Julien Abeille (jabeille)
Re: [Roll] RA-DIOs and next-door destinations Zach Shelby
Re: [Roll] RA-DIOs and next-door destinations Pascal Thubert (pthubert)
Re: [Roll] RA-DIOs and next-door destinations Jonathan Hui
Re: [Roll] RA-DIOs and next-door destinations Pascal Thubert (pthubert)