Geoff on IPv4 Exhaustion

Ted Mittelstaedt tedm at ipinc.net
Wed Nov 16 10:26:40 CET 2011


On 11/16/2011 12:52 AM, Shane Kerr wrote:
> Francois,
>
> On Tue, 2011-11-15 at 17:30 +0100, Francois Tigeot wrote:
>
>> It's ridiculous to not be able to configure hosts the same way it has been
>> done in IPv4 for years.
>> The bare minimum includes sending router and DNS information on a host-by-host
>> (not complete LAN segment) basis.
>
> Unfortunately the DHCPv6 standards do not yet support sending router
> information. This comes from ideological opposition to DHCPv6 within the
> IETF,

IPv6 isn't IPv4.  If you actually deploy IPv6 you will quickly find that
there isn't a need for DHCPv6 to send router information.

Consider an org that has a central site and several satellite sites 
under IPv6.  Remember, under IPv6 there is no NATTing so no need for
VPNs.  Instead the central site and satellite sites all have good 
firewalls.  Those firewalls know all about the IPv6 address information
at their site since they are getting it from the ISP they are using. 
But the network admin at the central site wants to distribute info
such as internal DNS servers, node types, and so on in a controlled
Microsoft domain.  He can setup a central DHCPv6 server for that and 
allow the routers at each site to handle the IP address assignment.
Under the old way there's DHCP v4 servers at every site that hand out
DNS info.  Under IPv6 there is only one, so if he makes a change in
that once then it goes to all remote sites, without having to login to
each site's firewall and change settings.

The problem here is not that you would want the capability of sending
router info in a DHCPv6 response to a LAN.  The issue is mainly one of
DSL and cable operators who have cudgeled DHCP v4 into an address 
assignment protocol for DSL and Cable modems that are basically behind
what acts like a point-to-point link.  They want to do the same for
IPv6.


Ted


More information about the ipv6-ops mailing list