Re: [Autoconf] new charter

[Alexandru Petrescu <[e-mail]>]

Stan Ratliff (sratliff) a écrit :
> 
> 
>> -----Original Message----- From: Alexandru Petrescu 
>> [mailto:[e-mail]] Sent: Friday, February 27, 
>> 2009 3:55 PM To: Rex Buddenberg Cc: Stan Ratliff (sratliff); 
>> Alexandru Petrescu; [e-mail] Subject: Re: [Autoconf] new 
>> charter
>> 
>> Rex Buddenberg a écrit :
>>> Stan,
>>> 
>>> The difference is not geo footprint (you got that right).
>> Rather the
>>> difference is between LAN (at fringe of network) and WAN (in the 
>>> interior).  A WAN will always be at least one router away
>> from end systems.
>>> While we're at it, references to SSIDs is not proper -- that's 
>>> 802.11-specific.
>> 802.11-specific because that's what many people actually mean. 
>> "Ad-hoc" type of SSID is there for same reason.  Many MANET 
>> experiments happened with 802.11 links.  Wireless LANs, WaveLAN, 
>> all are in scope and deserve mentioning.  They're the building 
>> blocks out of which MANETs can come up.
>> 
>> For LOS satcom communications: if I knew the precise names then I 
>> could find the manner in which they (or their admins) configure 
>> their IPv6 addresses.
>> 
>> I'm specifically keeping out of MANET 802.15.4 (akin to WPAN) 
>> because there seems to be an IPv6 ND draft for this link which 
>> already has its own address autoconfiguration quirks.
>> 
>> For WMAN: I'm aware of 802.16 and it has its own already IETF 
>> specified means of forming IPv6 addresses.
>> 
>> In this landscape, I'm not sure how could Autoconf practical way of
>>  forming addresses not mention WiFi.
>> 
>> Alex
>> 
> 
> Alex,
> 
> None of the MANET networks I deploy are based on WiFi. I think that 
> turning the discussion into one that is 802.11-centric is not a good 
> idea -- it's analogous to our earlier discussion on distance of the 
> radio link. Why should the Layer 2 protocol in use effect the 
> selection of Layer 3 addresses?

If not 802.11 then what link layer name?

I agree it's a good direction to stay independent on the link-layer name
- IP should run over them all.  At the same time I'm afraid once we
spell out the link layer name we implicitely name the way in which the
initial assignment of an IPv6 address to the nodes attached to that link
layer, is spelled out.  And we should at least reuse that.

Thank you for the oppinion, and I think many people think along the
lines you suggest.

Alex