Op 2 feb. 2014, om 06:41 heeft Lorenzo Colitti <[email protected]> het volgende geschreven:
> On Sat, Feb 1, 2014 at 2:45 AM, Teco Boot <[email protected]> wrote: > >> First we have to verify if this can happen. My favorite is using DHCP-PD > >> with server on CPE (edge) box (and elected box for ULA). This box should > >> circumvent your scenario. > >> > > Er, *what*? You are aware that if you have a partition, one of the two > > partitions will not be able to reach the DHCP server? > > Does it matter? Prefixes for unreachable DHCP servers should be deprecated. > Leases wil expire. A mobility tool is needed for connection survival (MPTCP, > whatever). > > You can't reliably deprecate information in DHCP. Even if the client supports > the reconfigure option, deprecating something in DHCP requires that the > server that has the original nonce still be up and around. > > This is why DHCP is not a good protocol to use in a distributed, unreliable > network like a homenet. Enterprise networks transform to distributed, unreliable networks (wireless access, BYOD with cellular access, sensors). At least some of them. Replace DHCP? I do not suggest full end-to-end DHCP connections from hosts to single DHCP server. I suggest structuring the Homenet somewhat, with clearly defined services for different tasks. I don’t see a problem using DHCP server on CPE box managing info for ISP connected via that box. And a DHCP server on leaf routers managing info provided to attached clients. Why not using DHCP between? Your reachability dependency argument is solved easily, let’s mandate the Homenet DHCP server advertises its DHCP server address (/128) in the routing protocol. Homenet routers watch these entries. Why not let Homenet routers MUST implement DHCP reconfigure extension, at least for their DHCP Client function? Of course I suggest DHCP as just one of the Homenet components. Teco _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
