Re: [Autoconf] Updated ad hoc addressing model document

[Thomas Heide Clausen <[e-mail]>]

On 4 févr. 2010, at 15:22, "Teco Boot" <[e-mail]> wrote:

> Hi Thomas,
>
>> -----Oorspronkelijk bericht-----
>> Van: Thomas Heide Clausen [mailto:[e-mail]]
>> Verzonden: donderdag 4 februari 2010 13:43
>> Aan: Teco Boot
>> CC: 'Charles E. Perkins'; [e-mail]
>> Onderwerp: Re: [Autoconf] Updated ad hoc addressing model document
>>
>>
>> On Jan 30, 2010, at 12:22 PM, Teco Boot wrote:
>>
>>> Charlie,
>>>
>>>>>>> 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 don't see that the addressing model prevents any such L3
>>>>>> communication.
>>>>>>
>>>>> How are packets forwarded?
>>>>> The destination address (which is direct L2 neighbor in this case)
>>>>> needs to be found in a forwarding table, normally the routing  
>>>>> table.
>>>>> Neighbor cache could be used also.
>>>>> How to get this info in such a table?
>>>>>
>>>>
>>>> Node A broadcasts a "Hello" message.
>>>> Node B hears it, and puts node A in its forwarding table.
>>>>
>>>> The nodes may take subsequent actions to verify
>>>> bidirectionality, exchange other table entries, etc.
>>>
>>> My point is that L3 communication becomes dependent on a L3 routing
>>> protocol. We didn't have this in the IP stack before.
>>
>> Well.....L3 communication depends on a populated routing table, thus
>> on something populating the routing table.
>> L3 multi-hop communications depends on something clever (a routing
>> protocol, a DHCP server, a human, for example and depending on the
>> place) populating routing tables.
>>
>> I do not see what this document does that changes that?
>
> Up to now, all IP addressing models I am aware of provide 1-hop L3
> communication between nodes that have L2 connectivity.
>
> The proposed addressing model for MANETs breaks this. Now there
> is a need for something clever. This clever thing could be
> stopped.
>


What makes the route to the 'local network' appear in the routing  
table? In that case the 'something clever' is whatever enters that  
route. Often, that 'something clever' is the same thing as what  
configures the interface....

Thomas

>
>
>>>> One should think of updating the IPv6 RFCs, since the link-local
>>>> constructions contained therein were written _explicitly_ in
>>>> disregard of the needs for wireless links of the sort familiar to
>>>> practitioners in this group.
>>>
>>> I use link-locals on wireless links, and it works great.
>>> What is your point?
>>
>> I am not Charlie, so I won't speak to what his point is.....but a
>> point can be made that you have observed a particular and specific
>> setup in which this holds true. That's good!
>>
>> This document aims at describing something that can hold in more than
>> one particular  and specific setup and, if respected, will allow
>> unencumbered operation, outside of a particular and specific setup.
>
> There is no reason not using link-locals for 1-hop traffic.
> If you say this is a particular and specific setup, is this a
> qualification for IPv6?
> If not, what are you referring to?
>
>
> Regards, Teco.
>
>