Re: RFC 5006 status

[james woodyatt <jhw@apple.com>]

On Mar 18, 2010, at 07:00, Ralph Droms wrote:
> 
> Stateless DHCPv6, in combination with RA/SLAAC address assignment, is simply not complex.  It does not require a full DHCPv6 server, easily fits in resource constrained devices and can be configured automatically.
> 
> Stateless DHCPv6 is certainly less complex than the DHCPv4 servers currently deployed in IPv4 gateways.

This has been my experience.

Configuring a stateless DHCP6 server is pretty easy compared to a stateful DHCP4 server, where you're forced to manage address assignments as well.  If you've already inserted a DHCP4 server into your residential gateway, it isn't that hard to add a stateless DHCP6 server for just the IPv6 DNS server addresses.

If RFC 5006 were to be revised for Standards Track category, then one "improvement" we might consider is to require O=0 in router advertisements that contain RDNSS options.  This would go some way toward addressing the ambiguity problem, while still allowing routers to advertise O=0 and PIO with A=1 and expect that hosts that SLAAC their addresses to those prefixes will be able, at least, to run applications that require them to be able to resolve domain names.

The main reason I'm a proponent of RFC 5006 is because it makes the O bit in the router advertisement something that can usefully be set to zero.  I suppose another way forward would be to deprecate O=0 in router advertisements.


--
james woodyatt <jhw@apple.com>
member of technical staff, communications engineering