Ole Troan wrote:
> ...
> would you be happier if we renamed it to SNCP (Simple Node
> Configuration Protocol)? :-)

Actually, yes. Routers are not hosts, so configuring routers with what
is titled a host specific protocol will create more confusion than it is
worth.

>
> for me this boils down to picking a packet format on the wire. DHCP
> offers a more flexible option format, is more extensible, has a
> defined relay mechanism, offers a Reconfigure mechanism so it is
> possible to poke the requesting router.  I have no problem with using
> ICMP PD, my point is that as we move along it is going to look awfully
> much like DHCP.

If we are going to need a stateful protocol tied to an authentication
system, using the wire format and option richness of a widely deployed
protocol makes more sense than inventing a new one. At the same time,
going that route requires a serious look at the defined options to find
any potential security holes that may exist because there was a
historical assumption that the option applied to an end system not a
router. If a router requesting one of the existing options opens a big
hole, we really do need a new protocol.

Tony



--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to