I think we are in violent agreement. I don't like the
IP->MAC->Customer mapping, it is forgeable, but it is the only one I
know we have available. I agree with you that it is not the only
possible mapping. If you can point me to a better existing mechanism,
I would be grateful.
Since DHCP maps a MAC to and IP address but not to a user, maybe there is a
technology that would provide this missing piece. Low and behold, there is
one: PPP and more specifically PPPoE (Over Ethernet) provides just the
functionality that some people are looking for. A number of broadband
providers are using PPPoE to fill the need of not only assigning an ip
address to a connection but associating that connection with a user.
Note the key phrase "better mechanism" in my statement.
Many people consider PPPoE an inferior technology. It reduces
reliability and usability of the product while raising the cost by
increasing complexity and placing unacceptable restrictions on the
PPPoE adds an unnecessary layer on the network stack. Thus it
consumes extra resources and gives one more component to fail and
PPPoE used as an authentication mechanism negates the "aways on"
advantage of IPoE networks. I don't like it when vendor engineers
impose restrictions on customers to satisfy the engineers design
problems. Good engineers design systems that make the network simple,
easy to use, secure, cheap and end-to-end. They don't come up with
ways to deliberately degrade the value of the network.
Personally, I think PPPoE belongs in the "bad idea box" along with
IPoATM and MPLS.
I would propose that if DHCP (or any other technology) does not do what you
want, see if there is another that will. If there is not build your own and
don't whine about it. Since the ISC DHCPD is open source, you can change it
as you see fit. If you change it, you should contribute back. However, do
not assume that Ted or anyone else will do this for you or support it if you
Hum, you must be thinking of someone else. Properly designed systems
based on DHCP do exactly what I want. My complaint is with people who
don't design networks properly.
Economic, Fast, Secure, Reliable and easy-to-use: If you can't design
a network system that is has all five qualities, best copy the system
of someone who can. Copying other's work until you reach master level
is a good thing.
I do so appreciate your suggestions about how open software projects
work. Lots of people think that because I have been doing this for so
long, I don't need these little reminders. But was one gets older,
the details slip out of your mind, like sea water running through the
fingers of your cupped hands.
Time for a walk on the beach.