RE: IPv6 policy at LACNIC( was: RE: [GLOBAL-V6]IPv6 Policy Proposal for LACNIC Region

["Michel Py" <michel@arneill-py.sacramento.ca.us>]

> Gert Doering wrote:
> Indeed there are two different goals colliding here.
> - reduce the IPv6 DFZ by all means, and whatever it might cost
> - get people to adopt IPv6 at all.
> The second goal means "as long as there is no other workable
> IPv6 multihoming solution, longer-prefix-assignment will be
> used" (or certain classes of early adopters won't adopt IPv6
> at all).

Indeed. Howewer, the second goal is likely to backfire, for two reasons:
1. Even if a policy common to all RIRs was to allow announcement of long
prefixes, multihomers won't buy it because they know that some people
will filter /48s no matter what the policy says.
2. Since the IETF has not produced anything in 10 years, it can be
doubted than the silver bullet will be found, therefore why bother with
IPv6 in the first place if it's to create the same mess as v4.


> The IPv6 DFZ will be smaller than the IPv4 DFZ anyway
> if people stick to "no aggregateable more-specifics
> announced from the same origin AS" (because it means
> "usuall every AS out there will announce only one
> prefix").

Although this is true, the actual gain remains to be seen. If indeed
multiple allocations to the same entity small bit by small bit are one
of the root causes of fragmentation, let's not forget mergers and
acquisitions, which often result in two merged ASes becoming one
announcing two prefixes. So in the beginning, a new address space would
be cleaner, but give it a few years....

Bottom line is, it costs less to defragment the IPv4 space than it does
to deploy IPv6 (remember that all the apps have to be re-written).

Michel.