Re: [Autoconf] Updated ad hoc addressing model document

["Teco Boot" <[e-mail]>]

Hi Thomas,

>-----Oorspronkelijk bericht-----
>Van: Thomas Heide Clausen [mailto:[e-mail]]
>Verzonden: donderdag 4 februari 2010 13:43
>Aan: Teco Boot
>CC: 'Ryuji Wakikawa'; [e-mail]
>Onderwerp: Re: [Autoconf] Updated ad hoc addressing model document
>
>Dear Teco,
>
>Thanks for your review. See comments below.
>
>On Jan 29, 2010, at 17:13 PM, Teco Boot wrote:
>
>> Ryuji, Thomas,
>>
>> I commented on the document.
>> I don't see any reflection in the document, nor received questions
>> for clarification.
>>
>> I am quite uncomfortable with a large drawback of the proposed
>> addressing model, which makes it unacceptable for the deployed MANETs
>> I am involved in.
>> My requirement is that L3 communication between nodes, that have L2
>> connectivity, must be possible in all conditions, including conditions
>> with a non-operational MANET protocol.
>
>I would wonder if you have a MANET if you are not running a MANET
>protocol?

The MANET protocol could get stopped, e.g. maintenance.
I am involved in MANETs in real world deployments. The requirement
comes from operational needs.


>That said, I am not sure I understand what the drawbacks you identify
>are. The document takes the "most conservative" approach, i.e. a
>network in which interfaces are configured in accordance to this,
>should allow any operation. The document, as I read it, uses "should",
>which does not prohibit alternatives (with the usual caveat concerning
>a "should").

I thought we should work on a practical addressing model.
In practice, it is useful to have IP access to nearby nodes,
also when the MANET protocol is non-operational.


>I believe that if you have no "MANET protocol", but still want L3
>communication between identified interfaces (IP addresses), then you
>would want a mechanism/protocol assigning these addresses? 

I don't want "no MANET protocol".
I want L3 connectivity when there is a L2 link, and the MANET protocol
is non-operational.


>           For the
>reasons outlined in that document, those addresses should (to allow
>any operation / any protocol to operate) satisfy the suggested rules
>in the document. If you do deviate from the "should", the usual
>caveats for a "should" apply -- and it might be OK for your deployment?

Yes, we all are free to ignore the document.
But there were strong opinions that the defined addressing models shall 
work in all scenarios?

The addressing model I use works great for all proactive MANET protocols
I am aware of. 
For the reactive routing protocols, there are some implementation issues:
The RREQ should be triggered after "address not reachable", i.e. interaction
with ARP/ND.



Regards, Teco