Great post. Everyone should read carefully what David said. This doesn't have to be so hard for EIDs. And the RIRs can still keep their jobs and CAN STILL ADD VALUE.
Dino On Mar 2, 2013, at 11:19 AM, David Conrad <[email protected]> wrote: > Scott, > > On Mar 2, 2013, at 8:43 AM, Scott Brim <[email protected]> wrote: >> EID space is not globally routed (or routable, which is different) but >> it is globally allocated, so you need some structure to manage >> allocation, including universally agreed on policies for initial >> allocations, further allocations, revocation, and more. Is there a >> better organization than the RIRs for all that? > > Yes. > > EIDs are globally allocated, why would you want to regionalize them? > >> The activities around >> routing are derivative from the fundamental reasons for the RIRs' >> existence. "Regional Internet Registries (RIRs) are nonprofit >> corporations that administer and register Internet Protocol (IP) address >> space and Autonomous System (AS) numbers within a defined region. RIRs >> also work together on joint projects." > > The RIRs are relatively heavyweight and increasingly political structures > that among other things, spend a lot of time trying to ensure the address > space they allocate can be routed. In the case of EIDs where routability is > not an issue, most of the policies and processes by which the RIRs provide > their services do not apply. For example, IPv6 addresses are (at least in > theory) allocated sparsely via a binary split-type algorithm to facilitate > growth of blocks while limiting the number of prefixes allocated. With EIDs, > you don't have to care about that. Another example: the policies for address > allocation varies per region and there are quite elaborate mechanisms in each > region by which those policies are created and modified. All of that > overhead would presumably be unnecessary with EIDs. Further, having > different policies for EID allocation based on arbitrary geopolitical lines > probably makes even less sense for EIDs than it does for IPv6 addresses. > > I suspect a better model for EIDs would be a simple, mostly automated > service, similar (not identical) to PEN/OID allocation done by IANA > (http://pen.iana.org/pen/PenApplication.page). > > Regards, > -drc > > _______________________________________________ > lisp mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/lisp _______________________________________________ lisp mailing list [email protected] https://www.ietf.org/mailman/listinfo/lisp
