Re: [dhcwg] dhc wg last call on agentopt-delegateand srsn-optiondrafts

On Fri, Jan 05, 2007 at 09:30:27AM -0500, Bernie Volz (volz) wrote:
> Unless you have the DHCP server itself participate in the routing
> protocols, you really have no other options?

I'm not aware of a better alternative, no.  Having the clients
perform the injection of the route is tantalizing, but unrealistic
with the current tools available.

That does leave only the DHCPv6 server as the last remaining party
to perform the injection, and as I think I detailed in past
exchanges I would have preferred my own network behave that way.

But as you and others have pointed out, that will not work so long
as the last leg between your final "router (really)" and "DHCPv6
client" is served by a device that is "a router that doesn't
participate in local dynamic routing protocols."  Eg the "i"
variety of BGP.

> If you were designing this, what would you do?

Probably go insane!

The very requirements you are working with, Bernie, are to effect
dynamic routing in a network without using any heretofore widely
tested and deployed dynamic routing protocol.  It is to me,
madness.  I would not come out of that sane, and it is a
statement to your endurance that you at least appear to have.

It is very much like asking someone to drink water from a tap
without using any of the array of glasses locked behind the
counter, but rather only what you have on your person as you
approached the bar.

I believe our species did not die of thirst before we learned how
to mold glass.  So too did we route.  I seem to recall living
through that era of mankind's digital evolution.  But I also
don't think we would consider the suggestion that cupped hands
might be used to intake liquids to be a keen idea all around,
now that we've already experienced more modern contrivances.

So too do I look with great suspicion upon this architecture
and design.  As a dynamic routing protocol, I find it not
merely wanting, but reminiscent of historic devices that caused
me, personally, ceaseless pain.  The idea of operating a network
with a dynamic routing protocol that looks remarkably similar
to RIP is no more appealing to me than drinking coffee with
cupped hands.

But really what we should talk about is that the problem your
draft solves is symptomatic of a deeper problem.

Which is the existence of "routers" that are beneath the task of
participating in the local routing protocols, but above the task
of merely operating their l2 magics.

Obviously I think there are at least two solutions to that problem,
one of which I might prefer...but I don't wield any power to
influence it, nor do I think any of the rest of us can.

And so I abstain (more vocally than I would have liked).

-- 
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