On 02/10/2014 13:26, Mark Townsley wrote:
> On Oct 1, 2014, at 9:44 PM, Brian E Carpenter <[email protected]>
> wrote:
>> 1) One reason for not stating homenet as part of the scope is
>> that we do not want to interfere with the current progress in
>> homenet. Personally I think there is a lot to learn from
>> homenet, but as I just said to Pierre, we are too late to affect
>> homenet's choices. I will be delighted if the results can be
>> applied to homenets in future, of course.
>
>> If we were having this discussion 5 years ago, I would agree.
>> But you homenet guys are ahead of us.
>
>
> Yes and no.
>
> Yes, homenet is ahead of anima in terms of, say, a distributed IPv6 prefix
> configuration algorithm. This is one of the first things the group began
> tackling, so there's quite a bit of water under the bridge here. However,
> while I have seen a lot of recent effort in security, homenet has a long way
> to go here. This happens to be something I get the impression anima has been
> working on for quite a while.
>
> You say that you wish to learn from what homenet has done, yet the current
> proposed anima charter says:
>
> ...autonomic service agents will demonstrate the usage of the above
> mentioned autonomic infrastructure components with two use cases:
>
> o A solution for distributed IPv6 prefix management within a network.
> Although prefix delegation is currently supported, it relies on human
> action to subdivide and assign prefixes according to local requirements,
> and this process could become autonomic.
>
> This use case is precisely what draft-ietf-homenet-prefix-assignment does
> (which has roots all the way back to draft-arkko-homenet-prefix-assignment-00
> in October 2011). So to homenet, this is a solved problem - with an algorithm
> that has been applied not just to HNCP, but to OSPF and ISIS.
Well, we have a bug in our short description, because the intention is to
support prefix assignment in a carrier scenario, which is different
in many ways. Good catch.
> I do think that there is room for a non-distributed algorithm that is tied
> more to centralized mechanisms, particularly as you move closer to a more
> tightly managed system. But for a distributed approach, as you observed
> Brian, homenet is rather far along.
> This is just the most obvious example that jumps out at me. There may be
> something similar to say about HNCP itself, the use of src+dst routing, etc.
> In any case, It's not hard to extrapolate from here that in a year's time or
> so, if we continue on the current trajectory, homenet will have come up with
> its own non-anima secure bootstrapping, and anima will have come up with its
> own non-homenet distributed IPv6 prefix configuration.
Which we should try to coordinate, since I see no reason in
theory why there can't be common underlying mechanisms between
enterprise, carrier and SOHO. But I don't want to hear in 2 years
time that homenet is stuck because anima hasn't met its milestones.
Brian
_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet
