Re: [Roll] FW: New Version Notification fordraft-thubert-roll-fundamentals-00

["Pascal Thubert (pthubert)" <[e-mail]>]

Hi Teco,

Please note that the tree formation in draft 01 incorporates your
suggestion (Richard had the same) about allowing to attach to a deeper
parent if the sequence is newer :) 

Please see below:

>For multiple trees (DAGs) and efficient routing without default routes
/ MTR
>/ flow labels, you could check Border Router Discovery Protocol and
related
>BRDP based routing:
>http://tools.ietf.org/html/draft-boot-autoconf-brdp
>http://tools.ietf.org/html/draft-boot-brdp-based-routing
>
>BRDP provides DAGs to attachment points, called Border Router. There is
no
>TIO (no Tree) but BRIO (Border Router).

Makes sense. Note that for 6LoWPAN backbone, the backbone owns the
prefix and is the virtual root of the aggregated tree.

>The sub-DAG tree identifier is the Border Router address. Nodes can use
>source addresses based on advertized BR prefixes. This enables using
these
>addresses as locators.


Added something similar in 01 from (again) Richard's suggestion

>BRDP / BRDP based routing use a routing protocol for reachability to
the
>backbone: next-hop to the Border Router. BRDP has its history in MANET
/
>Autoconf. It is easy to add paths to the backbone, learned from BRDP
(path
>to BR) or towards nodes (explicit messages, routing header and / or
>snooping). Routing to the backbone uses the source addresses, when
>destination is not in routing table. The requirement is that nexthop
towards
>BR is in the routing table.

As you know I like explicit messages but all 3 types of methods to route
towards node seem to have their own support. What's not totally clear to
me in the BRDP draft is the link / dependence to a MANET routing
protocol. Is there any? I mean when you say MANET router could we try
LLN router and get the same result?
 
>Check also loop prevention, I use this for building DAGs for BRIO
>distribution. It can be reused for loop-free routing. BRDP uses
hop-count
>and metrics, with seq.no. and thresholds for loop prevention.
Will do. You know that I do not like using UPM for loop prevention and I
prefer the depth.

The depth is a guaranteed Esperanto whereas the metric might be hardly
comparable over heterogeneous hops.

Also the roll-fundamentals draft enables forwarding along the best
metric while keeping the tree intact for a while, so we manage to get
similar benefit in terms of path metrics with a better topological
stability while using common sense depth as the base loop avoidance
method.

>For now, BRDP is IPv6. I planned to add IPv4 support in next version.
>Address generation is not possible (possibility for duplicates to large
for
>skipping DAD). Routing v4 over the DAGs (including SA based routing to
>backbone) is no problem.
>
>I have some ideas on using BRIO information for SA selection. Each node
has
>information on the best path to the backbone (or Internet default free
>zone), and a corresponding SA should be used for nodes not listed in
the
>routing table.
Makes sense

Cheers,

Pascal

>
>
>|-----Oorspronkelijk bericht-----
>|Van: [e-mail] [mailto:[e-mail]] Namens
Richard
>|Kelsey
>|Verzonden: dinsdag 7 april 2009 18:17
>|Aan: [e-mail]
>|CC: [e-mail]
>|Onderwerp: Re: [Roll] FW: New Version Notification fordraft-thubert-
>|roll-fundamentals-00
>|
>|   Date: Tue, 7 Apr 2009 17:53:35 +0200
>|   From: "Pascal Thubert (pthubert)" <[e-mail]>
>|
>|   I'm working on changes to accommodate multiple trees. The trouble I
>|   have is for the default route.  Whether the tree is Grounded or
>|   not, there can be only one tree used for default route, unless you
>|   enter MTR and flow labels. But the default is useful even if the
>|   tree is Floating.
>|
>|Makes sense.
>|
>|   The way I see it, we end up with a new bit, "default" and a node
>|   can only belong to one default tree. It'd better select a Grounded
>|   one. If a node belongs to 2 trees where the parent advertise
>|   default, the router will have to abandon the default on one of them
>|   as it propagates the TIO.
>|
>|   I think we can easily describe that but I'm unsure whether that
>|   goes in the main text or in 6.5.
>|
>|   What do you think?
>|
>|Moving between trees is discussed in item 6 in 2.1, so I
>|think it needs to mentioned there.  Perhaps that item should
>|be moved to 6.5?
>|                                    -Richard Kelsey
>|
>|----------------
>|This message and the information it contains are the proprietary
>|and confidential property of Ember Corporation and may be privileged.
>|If you are not the intended recipient, please do not read, copy,
>|disclose or distribute its contents to any party, and notify the
>|sender immediately.
>|_______________________________________________
>|Roll mailing list
>|[e-mail]
>|https://www.ietf.org/mailman/listinfo/roll