Re: [nemo] Multihoming and Load sharing.

[William D Ivancic <[e-mail]>]

Hello Theirry,

Actually most if not all of the multihoming documents refer to load sharing 
a policy routing.

draft-charbon-nemo-multihoming-evaluation-00
draft-ng-nemo-multihoming-issues-02
draft-paik-nemo-multihoming-problem-00

I agree that load sharing is not a MUST.  However it certainly may be 
desirable as is policy-based routing.   I simply want to point out that one 
should be careful to use such mechanisms wisely least your performance may 
suffer (or at least not gain as much benefit as thought) and/or you budget 
may suffer.  The later can be much more painful.

Will



At 04:45 AM 11/11/2003 +0900, Thierry Ernst wrote:

>Hi Will,
>
>When speaking about load-sharing, do you refer to a particular document
>?
>
>As far I know, load-sharing is only described as a potential benefit,
>not as something that must be supported. This is a result of being
>multihomed, not as pre-requisite. That fully allow it, other mechanisms
>are necessary on top of the signaling mechanism. The former may not be
>an IETF task.
>
>Thierry.
>
> > Whereas I can see configurations where it would be advantageous to allow
> > load sharing such as if one has two or three G3 wireless links up all with
> > relatively the same cost and bandwidth.
> >
> > However, IMHO there should definitely be a mechanism to turn load sharing
> > off.  It would be quite useless and very expensive to load share between
> > and WiFi link cost $80.00 per month unlimited service  and a 64 kbps
> > satellite link costing $1.00 per minute.