b) Is IPv6 good enough yet ?
I think so. There are valid concerns on two things; multihoming and
address allocation procedures. There seems to be strong forces among the
researchers and vendors advocting that we halt and wait for the
Grail of multihoming while we at the same time work very hard in preventing
people from getting allocations, to preserve address space and keep
the routing table size down.
A very good property of IPv6 is that we get to avoid some of the mistakes that were made with IPv4. One of those mistakes was giving out addresses in ways that didn't scale. History teaches us that once something like this happens the cats don't want to be herded back into the bag, to abuse some metaphors here.
I think the address allocation problem is based on v4 habits. v6 is abundant. We need it to be available. Nothing will happen until people can get real allocations with relative ease, not so easy that nuisance allocations will occur, but almost. Give every AS number holder a /32 or something, and watch deployment speed up.
There are now less than 35000 free AS numbers. If such a policy would be adopted there would be a huge land rush, depleting the AS number supply and forever polluting the IPv6 routing table with 64000 or so routes, most of which don't need to be there. The fact that this also uses up 0.0016 percent of the IPv6 address space isn't a significant issue, of course.
"Small ASes (those who originate only a few prefixes into
the global routing system) do not contribute more than their
fair share of either route entries or churn to the global
routing system."
Thus, if most ASen were able to contain all their hosts within one single /32 per AS, we would see limited routing table growth for some years, during which there would be time to develop more sophisticated routing paradigms, if operational experience dictated such a need was indeed present.
In IPv4 today multihoming is throttled by lack of routable address space. Without a /24 you don't get to multihome, and with a /24 - /21 you're never sure if you're going to be filtered. In IPv6 those would all be /48s that everyone can get so it's likely that growth in multihoming is going to be higher than what we've seen before. Also note that just keeping it below the level of Moore's Law (60% per year) isn't enough as routing table processing scales worse than linear at O = n.log(n)
Iljitsch van Beijnum
-------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
