Re: [Pppext] I-D Action:draft-hu-pppext-ipv6cp-requirements-00.txt

[James Carlson <[e-mail]>]

On 10/24/10 11:17 AM, Jacni Qin wrote:
> Dear John,
> 
> Thanks for your comments, please see below,
> 
> On Thu, Oct 21, 2010 at 2:48 AM, John Fitzgibbon <[e-mail]
> <mailto:[e-mail]>> wrote:
>     Bottom line: PPP is designed for establishing a link, DHCP is
>     designed for
>     (extensible) configuration, and Stateless Auto-Config is designed to
>     provide
>     basic IPv6 connectivity. Can't we just leave it at that?
> 
> 
> --> I'm not sure the if the consensus has been reached on the ND &
> DHCPv6 about this.
> Following your point, shall we propose to disable/get rid of IPv6CP,
> then run DHCP or ND over PPP?
> Or just because the interface-id thing concerns us that we can't do so?

The Interface ID problem is a fairly important issue.  IPv6's address
autoconfiguration mechanism depends on a source of IDs that's known to
be reasonably free of duplicates on a link.  The EUI-64 mechanism works
great on Ethernet, but isn't available on PPP, because PPP links just
don't have hardware addresses.

So, if one were to get rid of IPv6CP and just send unbidden IPv6
packets, how would non-duplicate IIDs be established?

In any event, I think this line of reasoning may be missing an important
point.  Even if the IID problem could be solved without IPv6CP, we have
the problem of existing implementations.

That trumps everything.  We have existing implementations of these
protocols, and they're deployed widely.  Making changes is no trivial
matter.  We can't just wipe the slate clean and do something arbitrarily
different.

I believe that includes adding extensions to IPv6CP to duplicate
functionality that's already deployed in DHCPv6.  And that's why I would
insist that we have a very compelling argument in favor of the proposed
drafts -- not just "this would be easier" but an explanation of why the
existing protocols in use today don't work.

-- 
James Carlson         42.703N 71.076W         <[e-mail]>