On Wed, Aug 06, 2008 at 01:30:05PM +1000, Richard Pruss wrote:
> Possibly this is the core of the disagreement.  I see it as critical to the 
> secure behaviour of an SP or Enterprise edge that the network 
> verify/enforce that the client is using the configuration the network gave 
> it.

I believe that is entirely true, and have been known to practice this
in my own networks with a zeal beyond that normally considered
practical or reasonable.

> I see it as DHCP's role to configure the host and the network at the 
> boundary between layer 2 and layer 3.

DHCP's role is to configure the host, yes, the network, no.

Unless the network elements were themselves DHCP clients.

The trouble with the relay's role in a DHCP host's exchange is that
it is not given a state.  This is wonderful, because it makes relays
very simple pieces of software, having no state to manage.  For folks
that are trying, as I said, to make DHCP do what it wasn't designed
for, it is an incomplete fit, and problems constantly arise with
out-of-order packet delivery, out-of-band packet delivery (where
there is a second relay), and inconsistent replies made to clients
(of which only one is accepted by the client) which are particularly
problematic when more than one server is involved.  To name just a
few.

Given time to sort it out, and neglecting pathological cases, a DHCP
server and DHCP client will converge on a single image of the desired
configuration.  A DHCP relay is left out of this convergence, so
you cannot reliably convey configuration.

I believe it simply was never a design criteria, and am more or less
at peace with that as the ultimate designs of both DHCP protocols
help them fit where I believe DHCP belongs.

Which doesn't include access control.

-- 
Ash bugud-gul durbatuluk agh burzum-ishi krimpatul.
Why settle for the lesser evil?  https://secure.isc.org/store/t-shirt/
-- 
David W. Hankins        "If you don't do it right the first time,
Software Engineer                    you'll just have to do it again."
Internet Systems Consortium, Inc.               -- Jack T. Hankins

Attachment: pgpT1bqaOxKGK.pgp
Description: PGP signature

_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to