Re: [v6ops] New Version Notification for draft-yourtchenko-ra-dhcpv6-comparison-00.txt (fwd)

[Ole Troan <[e-mail]>]

> can I paraphrase the problem?
> "the problem is how to get hosts to load balance across multiple first-hop routers?"
> 
> That's an easy problem to solve - do ECMP on the hosts.

right.

> Unfortunately, though, I fear that the problem is "we want the IPv6 design to be the same as the IPv4 design we already have", and in general that's a problem that's impossible to solve without making IPv6 look the same as IPv4.

that's not a class of problem I think we're able to solve in the IETF.

I think it is a perfectly valid operational model to manually / statically configure hosts though. Router Discovery provides a dynamic way to discover a router for default, that handles multiple routers, adjusts dynamically as the network changes.
manual configuration doesn't do any of that, but that doesn't mean there isn't a place for manual configuration.
if operators want to configure the default route with something like Puppet or Netconf isn't that acceptable with you?
DHCP would just be another way to do manual configuration of static information to hosts.

if this was implemented in DHCP, and the expectation was that all hosts should support it,
how does the deployment model and transition phases look like?
e.g. you have a mixed link with some hosts supporting the option and some hosts that don't.
is static DHCP configuration mutually exclusive with router discovery?

cheers,
Ole