Hi Mikael, See inline. On Jul 9, 2012, at 7:30 AM, Mikael Abrahamsson wrote:
> On Mon, 9 Jul 2012, Jari Arkko wrote: > >> Hi, >> >> We've submitted a new version of our OSPFv3-based prefix assignment draft. >> This version adds quite a bit of detail to the algorithm, describes ULA >> prefix generation, etc. >> >> Here's the link: >> >> http://tools.ietf.org/html/draft-arkko-homenet-prefix-assignment-02 >> >> Comments appreciated. > > Hi, > > I haven't really read up what has been discussed historically here, so > please have patience in case I am rehashing what has already been > discussed. > > From reading the text it is not clear to me exactly what deployment > scenario the text is meant to handle. Some ascii drawings might help? I've > been pondering how to solve a few different problems for deployment, for > instance the problem of the customer having a CPE with a live DHCPv6-PD > prefix, unplugging it by means of power-off, and then putting in another > CPE. If only a single prefix is allowed on the link, then the customer > would have to wait until the PD lease expires until the new CPE can get > one (the same problem exists for IPv4 as well, in case only a single IPv4 > address is allowed and the customer just powers off the CPE without doing > DHCP RELEASE first). > > I've been considering allowing 4 concurrent PD prefixes, for instance the > customer has a statically configured /56 as PD pool, a /64 on-link (which > is not from the /56 space), and then handing out /58 with PD from the > customer-unique /56. This would mean that the deployment scenario might in > worse case look like this: > > ISP handoff switch/router > |/64 > -----customer switch---------------- > | | | | | | > CPE1 CPE2 CPE3 CPE4 PC1 PC2 > /58PD /58PD /58PD /58PD > | | |---| > L2 CPE5 CPE6 CPE7 > | | | | > PC3 PC4 PC5 PC6 > > This is in a ETTH scenario, or could also be over for instance DSL where > the modem is a bridged L2 handoff. To solve the "power off" problem, I > guess it might work with a /57 per CPE as well, and hope the customer long > term only wants to have a single CPE. > > Are CPE1-4 going to talk OSPFv3 between them on the link there according > to the draft, or is this just for the CPE3/CPE6/CPE7 shared segment? Normally, an auto-configured OSPFv3 CPE will not run OSPFv3 on the link toward the ISP. However, the auto-detection is not standardized. Excerpted From the OSPFv3 auto-configration draft: 2. OSPFv3 SHOULD be auto-configured on for IPv6 on all interfaces intended as general IPv6-capable routers. Optionally, an interface MAY be excluded if it is clear that running OSPFv3 on the interface is not required. For example, if manual configuration or an other condition indicates that an interface is connected to an Internet Service Provider (ISP), there is typically no need to employ OSPv3. However, note that in many environments it can be useful to test whether an OSPFv3 adjacency can be established. In home networking environments, an interface where no OSPFv3 neighbors are found but a DHCP prefix can be acquired may be considered as an ISP interface. Thanks, Acee > > -- > Mikael Abrahamsson email: [email protected] > _______________________________________________ > homenet mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/homenet
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
