Nice migration to HOMENET. ;)

On 8/1/11 2:58 PM, "Ole Troan" <[email protected]> wrote:

>[v6ops in bcc]
>
>> On Sun, Jul 31, 2011 at 10:38, Brzozowski, John
>><[email protected]> wrote:
>> The outer most router supports DHCPv6 PD (LAN side).  It would have to
>>be
>> seeded with adequate address space to serve up IA_PD (and possibly
>>IA_NA)
>> for downstream routers.  Based on conversations to date I assume there
>> would be interest in downstream routers also support and being able to
>> delegate prefixes?
>> 
>> There are two issues I see with DHCPv6 PD to distribute prefixes inside
>>the home:
>> 
>> 1. Prefix delegation only really works in a tree topology, so when the
>>user plugs things in wrong (e.g., plugs a LAN port into another LAN
>>port, or creates a loop) it won't work. Having a routing protocol will
>>allow things to work regardless of how the user plugs things in.
>
>prefix delegation with DHCP can be made to work in arbitrary topologies.
[jjmb] agreed.
> 
>
>> 2. We already use PD between the home network and the ISP. If we use it
>>inside the home as well, then a home router getting a prefix via DHCPv6
>>PD will not know whether it is connected to the ISP or not, and thus if
>>it should turn on the firewall or not.
>
>PD (or not) as a mechanism to detect a network boundary is probably not
>reliable or secure enough.
>
>> Both of these issues can be solved by using an IGP.
>
>3. DHCP PD does not handle the multi-home / multi-prefix case. an IGP
>based solution would.
[jjmb] can you elaborate on this?  Why not?
>
>the requirement to have an IGP is somewhat orthogonal to the problem of
>prefix assignment. the autonomous properties of a flooding mechanism as
>opposed to a request/reply mechanism is the key in my opinion.
>
>cheers,
>Ole
>

_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet

Reply via email to