Re: Source address selection in IPv6 multihomed multi-addressed sites

["Kenji Ohira" <torus@yakitori.no-ip.com>]

Cedric,

----- Original Message ----- 
From: "Cedric de Launois" <delaunois@info.ucl.ac.be>
Subject: Source address selection in IPv6 multihomed multi-addressed sites

> I believe this problem needs to be resolved to get
> a complete IPv6 multi-addressing multihoming solution.
> We assume here that the providers perform ingress
> filtering.
> I'm here only looking to solutions where the source
> address selection is not made by applications, but it
> is done "automagically". However, for all these solutions,
> it is possible to imagine some mechanism where the
> applications can overload the automatic behaviour.
First of all, I think that source address selection should be made by each 
application/transport layer.

> Here is my understanding of the different solutions
> proposed.
> 
> * A solution is to perform source address based routing
>    inside the multihomed site. The packets are always
>    routed to the site exit router corresponding to the
>    source address used.
> 
>    Pros: - no new option/protocol
>          - no modification to hosts
>          - no tunnel
>          - routing efficiency (dumb routing) ?
> 
>    Cons: - possibly very bad load-balancing if all the
>            hosts choose the same source address (and this
>            is what they'll do with RFC3484 !) +
>            no way to engineer the traffic if no additionnal
>            mechanism is used
>          - source address based routing has probably
>            much more implications than what it is
>            expected. This needs to be studied further.
> 
>    My point of view : maybe source address based routing
>    is not something desirable. Anyway, source address based
>    routing alone is not sufficient.
We are interested in application-level load sharing more than that of site-level.
I think that it is related to QoS issue, too.
If an application wants to use all bandwidth of all external connection, it is able by 
selecting all address which the host has.
If QoS parameters of each path are differ, an application can select the best suited 
path which it needs.

If an application has no special requirements for any QoS parameters, the application
can leave this selection to transport layer protocol as like as SCTP.
In this case, at the timing of establishing a connection, transport layer protocols on both 
end hosts negotiate all addresses which each host has, and each host select the best pair 
of src/dst addresses in according to RTT.

As another solution, an application or a protocol of transport layer can select a path in 
according to the longest match between source address and destination address.
In order to talk about this, I talked about 3-level hierarchical network model.
As more other solution, an application or a protocol of transport layer can select route in 
according to the external database as like as NAROS.

Anyway, a selection of source address can be and should be resolved in layer 4 or above.
For this reason, IP layer should be simple as possible in order to select correct site-exit 
router which is corresponding to a source address selected in layer 4 or above.

July 17, 2003 (JST +0900)
----
Kenji Ohira
Graduate School of Informatics, Kyoto University, JAPAN
mailto: torus@tori.cc / ohira@net.ist.i.kyoto-u.ac.jp