Brian,

> It appears from the discussion that the "network administrator" is trying to 
> get *multiple* Linksys/equivalent systems to work together with no 
> intervention (and potentially with multiple, independent ISPs).  None of the 
> people who I know who have such a setup with IPv4 expect this to work "out of 
> the box" and that is what I see people trying to do here with ULAs.

not quite as complicated as that even. two CPEs routers side by side 
(presumably connected to different ISPs). if there is a requirement that a CPE 
router should automatically generate a ULA, should the 2 routers then 
coordinate the ULA assignment between them.

as you say, I'm not aware of any networks like that which you can 
auto-configure for IPv4 either. and the benefit of a single ULA prefix versus 
two when you in any case don't have zeroconf routing.

I'm trying to get an idea what IETF consensus is for these two BBF 
requirements. I take your opinion to be: this is not a problem we should solve 
(it really requires a lot of other things too).

cheers,
Ole


> Fred Baker wrote:
>> well, of course. The question isn't what the RFC was written for, it's what 
>> it might be used for. In this case, the "network administrator" is the 
>> person who in today's internet installs a Linksys/equivalent system in the 
>> residence/SOHO and expects to to work before they have attached to the ISP. 
>> It works with IPv4...
>> On Jan 15, 2010, at 9:43 AM, Brian Haberman wrote:
>>> Wojciech Dec (wdec) wrote:
>>>> In general, reading through the ULA rfc, while there is a fair bot of
>>>> talk regarding pseudo-random ULA global-id's and use along with SLAAC,
>>>> there hardly is any reference to the scenario where there can be
>>>> multiple global-id's per site sourced by multiple routers. However, the
>>>> presence of a subnet-id indicates that the authors did have in mind a
>>>> more managed addressing assignment regime, which becomes undone in the
>>>> multiple router/gateway case.
>>> 
>>> The ULA RFC was not written with the perspective that individual routers 
>>> would automatically generate the ULA prefix and then advertise them (either 
>>> in RAs or a routing protocol).  Rather, a network administrator would 
>>> generate the ULA prefix using the guidelines provided, design a subnet 
>>> model for the network, and then configure the ULA prefix + subnet 
>>> information in the routers.
>>> 
>>> If a network admin wanted multiple, diverse ULA prefixes, he/she can use 
>>> the random generation logic to generate an arbitrary number of them. Again, 
>>> the RFC was not written with the intent of routers automatically generating 
>>> the ULA prefix without operator intervention.
>>> 
>>> Regards,
>>> Brian
>>> 
>>> --------------------------------------------------------------------
>>> IETF IPv6 working group mailing list
>>> [email protected]
>>> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
>>> --------------------------------------------------------------------
>> http://www.ipinc.net/IPv4.GIF
> 
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> [email protected]
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------

--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to