To:
Tim Chown <tjc@ecs.soton.ac.uk>
Cc:
dnsop@cafax.se
From:
Alain Durand <Alain.Durand@Sun.COM>
Date:
Tue, 29 Jul 2003 13:19:38 -0700
In-reply-to:
<20030729191730.GE31891@login.ecs.soton.ac.uk>
Sender:
owner-dnsop@cafax.se
Subject:
Re: proposal for a compromise on DNS discovery
On Tuesday, July 29, 2003, at 12:17 PM, Tim Chown wrote: > On Tue, Jul 29, 2003 at 10:55:01AM -0700, Doug Barton wrote: >> >> 1. In every situation where RA DNS announcements are proposed, dhcp >> is a >> better solution. > > But will every device have a DHCPv6 (or DHCPv6 Lite) server on link? > For > those that do, one might ask "why bother with stateless autoconf?" and > of > course many sites will wish to keep using DHCP(v6) as a policy choice. > But for networks without DHCP on link, the RA method is an alternative. One could very well envision a network where the routers will act as DHCP-lite servers, only passing "stateless configuration" like DNS server, NTP server, .... but NOT allocating IPv6 addresses. >> 2. RA DNS announcements are an ugly feature creep for RA, and should >> not >> be implemented. > > The spec for Router Advertisements allows for additional options to be > specified, so this was considered possible when the spec was done. So the next step would be to define NTP server address in RA, web proxy address in RA, SIP server address in RA,.... RA could be extended to pass anything that could be a DHCP option. This is where feature creep happen. - Alain. #---------------------------------------------------------------------- # To unsubscribe, send a message to <dnsop-request@cafax.se>.