Re: [v4v6interim] You have been dugg!

If the DHCPv6 PD option (RFC3633) is used to assign a prefix to the CPE
router in a home, then sure, since the PD option format includes a
prefix length, by all means, any ISP can assign the CPE router a prefix
of length the ISP so desires by their network design.  I was talking
about DHCPv6 without the PD option where DHCPv6 address assignment does
not include a prefix length. Doling out a PD option to a router is
different from DHCPv6 address assignment. 

Hemant 

-----Original Message-----
From: Fred Baker (fred) 
Sent: Monday, October 13, 2008 6:48 PM
To: Hemant Singh (shemant)
Cc: Iljitsch van Beijnum; v4v6interim@ietf.org
Subject: Re: [v4v6interim] You have been dugg!

um, hold on. two uses.

DHCPv6 is used to assign host addresses to host interfaces. They only
have one prefix length: /128.

DHCPv6 is also used to assign a prefix to the CPE router in a home.  
the addcon draft, in large part, is saying that these are not always
(and perhaps should never be) /64. Please tell me that we can assign a
/60, a /56, or a /48 to a home if we want to.

On Oct 13, 2008, at 3:33 PM, Hemant Singh (shemant) wrote:

> I haven't had the time to catch up to these emails and get a proper 
> perspective.  But just a quick comment that DHCPv6 is not broken if
> DHCPv6 responses from the server to the client do not include prefix 
> length - that's by design of the IPv6 architecture where the router on

> the network is responsible for doling out the prefix length to the 
> clients and NOT the DHCPv6 server.
>
> Hemant
>
> -----Original Message-----
> From: v4v6interim-bounces@ietf.org 
> [mailto:v4v6interim-bounces@ietf.org
> ]
> On Behalf Of Fred Baker (fred)
> Sent: Monday, October 13, 2008 6:23 PM
> To: Iljitsch van Beijnum
> Cc: v4v6interim@ietf.org
> Subject: Re: [v4v6interim] You have been dugg!
>
>
> On Oct 13, 2008, at 5:42 AM, Iljitsch van Beijnum wrote:
>
>> On 7 okt 2008, at 19:26, Fred Baker wrote:
>>
>>> What does it mean to be "IPv6-like"?
>>
>> That everything happens mostly automatic and is easy otherwise. For 
>> instance, in IPv4 you have to think about how big each subnet is. In
>> IPv6 you don't.
>
> I disagree. The discussion of /48 vs /56 vs /60 vs /64 comes to mind.
> And btw this is about a IPv4 address, like NAT64 etc are. You make 
> relatively positive statements about the other options and dump on one

> that has the same strengths and weaknesses.
>
>> The IVI mapping means that at the very least a special subnet has to 
>> be created for every IPv4-reachable host. The suggestion of having
>> non-/64 subnets and DHCPv6 is very problematic because DHCPv6 doesn't

>> know about subnet prefix lengths.
>
>
> we have a way to assign a prefix using DHCP. You're telling me it 
> doesn't know the prefix length? If so, it's broken. I haven't read 
> that spec; remind me which it is?
>
> And yes, regardless of the address format used - even SLACK, for that 
> matter - one has to either inject a host route into routing or inject 
> a prefix into routing. That's how routing works. The only way we can 
> avoid that is to put every mapped-address host on a LAN shared with 
> the translator. Bzzt. That doesn't scale.
>
>> The situation where a static mapping is set up means that the 
>> configuration can be limited to the translation box, this simply 
>> makes
>
>> much more sense on every level.
>
> Actually, not so. And this differs from NAT64, which you spoke 
> positively about, in what way?
> _______________________________________________
> v4v6interim mailing list
> v4v6interim@ietf.org
> https://www.ietf.org/mailman/listinfo/v4v6interim

_______________________________________________
v4v6interim mailing list
v4v6interim@ietf.org
https://www.ietf.org/mailman/listinfo/v4v6interim