[ipv6hackers] "Stick to limited IPv6 deployments, businesses warned"

[Owen DeLong]

On Sep 6, 2012, at 00:03 , Sander Steffann <sander at steffann.nl> wrote:

> Hi Owen,
> 
>> SLAAC is useless in 99% of cases without RDNSS. If you have to implement DHCP, you might as well go all the way.
> 
> I don't agree at all here. SLAAC+Stateless DHCPv6 works fine. With RDNSS you can optimize a bit and avoid a few extra packets and a little bit of software (stateless DHCPv6 does not require that much code), but SLAAC+Stateless DHCPv6 is certainly not useless. It's what I use on most unmanaged networks.
> 

If I'm going to the trouble of configuring DHCP on my server/router/etc., then I might as well just configure DHCP and skip SLAAC.

Who cares how much code it is or isn't, it's already written and it's already part of the OS. What I care about is how much I have to configure.
Having to configure in multiple places just to get complete autoconf information working sucks. I'd like to be able to configure SLAAC+RDNSS on the interface of the router and be done in a lot of cases.

Yes, I can make DHCP work, but I'd rather not have to just to get DNS servers, especially since it DOESN'T work even for that with MacOS prior to Lion or Windows prior to Vista or default installs of most deployed versions of Linux/BSD, etc. on the client side. (Admittedly, neither does RDNSS, so this totally sucks no matter what).

All of this is the unfortunate result of the fact that instead of developing complete solutions, various factions in the IETF spent all their time making sure that the other side couldn't get to consensus on anything complete.

As a result, we have half-assed DHCP with no routing. We have half-assed stateless autoconf with DNS servers as an afterthought and we have this hybrid approach of using partial implementation of DHCP to overcome the shortcomings inherent in SLAAC without RDNSS.

What we need is to complete these solutions on all sides so operators can choose the set that best fits their environment.

Owen