On 2012-03-28 16:34, Cameron Byrne wrote: > On Mar 27, 2012 7:09 PM, "Brian E Carpenter" <[email protected]> > wrote: >> Dave, >> >> On 2012-03-28 09:28, Dave Taht wrote: >>> On Thu, Mar 22, 2012 at 3:50 PM, Brian E Carpenter < >>> [email protected]> wrote: >>> >>>> On 2012-03-22 12:33, homenet issue tracker wrote: >>>>> #4: Use of ULAs >>>>> >>>>> CN1 in the -02 text says ULAs should be provisioned by default. Do > we >>>>> agree? >>>> Yes, by the CER (MUST). >>>> >>>> It's much less clear for all other routers on site. I would prefer that >>>> no other router provisions a ULA prefix if the CER has already > delegated >>>> one. >>>> >>> Since it is nearly impossible to learn a subnet prefix that is unused >>> unless all devices share a routing protocol, I have generally preferred > to >>> generate unique ULAs per routing device. >> Our two statements are not inconsistent... obviously, if a router comes >> up and does not hear about a ULA prefix from another router, it may be >> appropriate to generate a new one. But if we have a prefix delegation >> mechanism in place, that may be sufficient. >> >>> >>>>> And if so, should they be preferred over globals? >>>> Yes. >>>> >>>> >>> This part I don't quite understand. I would prefer global prefixes to >>> always take precedence over ULAs. >>> I'd like ULAs to try to talk only to ULAs. ULAs should not escape into > the >>> global DNS, but should end up in local dns. >> ULAs should really only be used to talk to ULAs, so when I say 'prefer > ULAs', >> I mean prefer a pair of ULAs to a pair of globals. In case it isn't clear, >> in my mind the problem is always address *pair* selection. So that means >> that a host should do something like this in the default case: >> >> Do I have a ULA? >> If yes, does the other guy have a ULA (in the same prefix)? >> If yes, use the two ULAs. >> Else, use two globals. >> >> I'm hoping 3484bis will result in this. >> > > In the general case, you don't believe ula and global should have > communication at all? Or just a matter of source selection when multiple > address types are available to a given host ?
The second. If the only choice is ULA<->global, of course you try it, but then happy-eyeballs may kick in. Brian > > Cb > >>> Even with 'happy eyeballs', having a means to disable a ULA -> global >>> connection by default saves 100s of ms in the bad cases. >> Agreed. >> >> Brian >> >>>>> The new >>>>> 3484-bis has changed so they are not *unless* a specific /48 for the >>>> site >>>>> is added to the 3484 policy table with higher precedence than > globals. >>>> We >>>>> should design something that works when disconnected from the > Internet. >>>> We'll have to deal with whatever 3484bis ends up saying. If it does >>>> require an explicit policy table entry table, that will require >>>> a mechanism, associated with the prefix delegation mechanism. >>>> >>> OK, I will re-read, thx for the update.. >>> >>> >>>>> Also, we currently say nothing about ULA-only devices and their >>>>> reachability from outside the homenet; do we want to? >>>> I don't see what there is to say; they aren't reachable, and that's >>>> a feature. >>>> >>>> Brian >>>> _______________________________________________ >>>> homenet mailing list >>>> [email protected] >>>> https://www.ietf.org/mailman/listinfo/homenet >>>> >>> >>> >> _______________________________________________ >> homenet mailing list >> [email protected] >> https://www.ietf.org/mailman/listinfo/homenet > _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
