On Jun 2, 2013, at 11:10 AM, Ted Lemon <ted.le...@nominum.com> wrote:
> On Jun 2, 2013, at 11:59 AM, Owen DeLong <o...@delong.com> wrote: >> You are assuming that all of the subordinate routers will act as DHCP relays >> rather than doing PD. >> That is certainly one possible solution, but, not necessarily ideal in all >> cases. >> In cases where the subordinate routers should receive delegations and >> perform their own PD for their subordinate routers, having a larger bit >> field can be useful for greater flexibility. > > No, there is no use case where this is better than doing the delegations from > the router that received the initial delegation (since we're apparently just > arguing by vigorous assertion). > We can agree to disagree. One example that comes to mind is if I want greater control and I want my most capable router with the greatest configuration flexibility to be in charge of the addressing scheme, but, it is not the router that interfaces with my ISP. >> Thus, providing 16 bits to the end site is, IMHO, worth while. > > And hence, this conclusion is not supported. > > You are welcome, of course, to contradict me by stating such a use case, but > bear in mind that when you delegate prefixes for further sub-delegation, > topology changes in the homenet become impossible. So your use case for > doing this would have to enable some pretty awesome functionality before it > would be worth doing. Also make sure you think about how it would work > during a renumbering event, with sub-delegations and sub-sub-delegations all > having different lifetimes. > Actually, the need for the larger bit field is precisely to allow topology changes in said deployment scenario. If the top level router hands out, for example, /50s to its 3 subordinate routers, the subordinate routers can support a number of different topologies without requiring any changes at the top-level router. Additionally, a fourth subordinate router can be added with its own underlying topology supported. OTOH, if there are more than 3 subordinate routers, the top level router can delegate /51s. True, this would complicate the change from 3 to more than 3 subordinate routers at the top level somewhat. > (I've got nothing against delegating /48's to the home, but the reason we did > that was to maintain flexibility, not because we really expect a typical > homenet to have 65,536 subnets. At least for most reasonable values of > "we.") > You just said exactly what I said to begin with… It's to have a bit field wide enough to allow flexibility in the automation of the hierarchical assignments, not to create 65K subnets. I never asserted it was because we needed 65K subnets. Owen
-------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------