host-identifier with IPv6

[Frank Sweetser]

Ted Lemon wrote:
> On Mar 2, 2009, at 5:15 PM, Frank Sweetser wrote:
>> Sorry, but I've got several hundred lab machines that refute the idea 
>> that
>> Sten's scenario is contrived.
> 
> I don't mean "contrived" in the sense of "never happens in practice."   
> I mean contrived in the sense that using it to prove the point Sten was 
> trying to prove is contrived, because I can come up with an opposite 
> example that proves the opposite point, and is an equally valid 
> example.   Thus, the answer can be attained by following neither 
> example, but instead must account for both.

Fair enough.  Personally, I suspect that in most cases software changes would 
wipe the DUID far more frequently than hardware changes that would change the 
MAC address, but can I see your point.

> Believe me, I understand and sympathize.   I have to maintain machines 
> like this too.
> 
> But remember that DHCP deployments in the millions of addresses per 
> server are not uncommon in the ISP space, so although your 300 are 
> significant, there are many other situations out there that DHCP also 
> has to support.

And for those situations, I can see how adding the DUID value can solve 
problems.  Even on our comparatively tiny network with a few thousand hosts, 
we've twice run into duplicate MAC addresses from sleazy vendors.  (Of course, 
that broke ethernet altogether, since they were on the same LAN, so the fact 
that this broke DHCP didn't even enter into the picture for me =)

The kicker for me, then, isn't so much the DUID and the rules that go into 
generating it.  It's really the removal of a field explicitly defined as the 
client hardware address from the client requests that wipes out a lot of my 
current workflow, and, since those million client sites can freely ignore the 
client HW address in favor of the new DUID, its removal feels rather 
gratuitous to those of us who depend on it every day.

-- 
Frank Sweetser fs at wpi.edu  |  For every problem, there is a solution that
WPI Senior Network Engineer   |  is simple, elegant, and wrong. - HL Mencken
     GPG fingerprint = 6174 1257 129E 0D21 D8D4  E8A3 8E39 29E3 E2E8 8CEC