Re: [dhcwg] FW: New Version Notification for draft-wing-dhc-dns-reconfigure-01.txt

[Mikael Abrahamsson <[e-mail]>]

On Fri, 7 Jun 2013, Prashanth Patil (praspati) wrote:

> Hi all,
> This draft proposes a mechanism to extend DHCPv6 such that a DHCPv6 Relay
> Agent can dynamically influence priority of DNS servers provided to a
> host, so that a host can use an optimal DNS server for resolution.

I feel this proposal have two problems in my opinion:

I don't like "flags" that means the server responds differently with the 
same field. This means there must be server-side logic to handle the 
different cases. I believe other solutions which work with a static 
server-side configuration file without "if/then/else"-statements are 
preferred.

I don't like the chattyness of the proposed solution, where if a client 
goes between different states of dualstacked-ness, it needs to talk DHCP 
every time.

As I see it, it also doesn't address an additional problem, where if the 
client has gotten DNS resolvers from both DHCPv4 and DHCPv6, there 
currently is no sure way to know what the client will use (I know for a 
fact that different android phones from different vendors behave 
differently). If we're going to enhance this, it would be beneficial if 
this case would also be solved.

So I propose a change of the solution where the client would initially ask 
the following:

What resolver should I use if I am single stacked IPv6.
What resolver should I use if I am dual stacked, and also please tell me 
if there is a preference if this resolver is better, same or worse than 
any other resolver I might have gotten provisioned any other dynamic way 
(DHCPv4 comes off my mind).

I am not knowledable about DHCP to know what the best way to actually 
implement this, but from an operational point of view, I would really like 
to see the above goals be achieved if we're enhancing the DHCPv6 client 
and host stack behaviour anyway.

-- 
Mikael Abrahamsson    email: [e-mail]