SLAAC vs DHCPv6 (Re: RFC4941bis: consequences of many addresses for the network)

[Jared Mauch <[e-mail]>]

> On Jan 23, 2020, at 9:08 AM, Fernando Gont <[e-mail]> wrote:
> 
> * DHCPv6 support is not required, and in fact we seem to do our best to
>  trash DHCPv6 and borrow cool DHCPv6 features into slaac as much as
>  possible -- which doesn't help the situation.

This for me is an example of one of the problems around IPv6.

The IPv6 solution space includes things like:

* Don’t do DHCPv6, do SLAAC
** We forgot to include DNS, so RDNSS
* Do DHCPv6, don’t do SLAAC
** Some hosts won’t do DHCPv6
** Not all hardware does DHCPv6-PD properly
** Not all software does DHCPv6-PD properly
** Not all networks can support DHCPv6-PD

The general discussion for me here is we need to ensure we’re not violating rfc1925[1] 2.1 which at times seems to be blame shifted to operators or vendors.  While everyone has their weaknesses we need to ensure the ecosystem we specify is robust enough.

- Jared

[1] - Yes I know, doesn’t make it any less true though.