[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Draft on Globally Unique IPv6 Local Unicast Addresses

    Date:        Tue, 27 May 2003 09:40:53 +0200
    From:        Brian E Carpenter <brian@hursley.ibm.com>
    Message-ID:  <3ED31685.DAD1937B@hursley.ibm.com>

  | Let's not restart the ICANN wars here over a trivium.

I don't want to, but someone sure will.   This is exactly why the WG
declined to go down this path way way back when it was first being
discussed (you may remember that it was me who proposed this, a long
long time ago).

  | Like any other
  | registry, we make it unique and its authority would come from IANA.

Sure, we can do that, as we do for the root zones for the DNS.   But,
we know that has competitors - they exist, even though the names they
sell are useless for most practical purposes.

The numbers we're talking about would be just as useful as any others,
as they have no function beyond being numbers.   Anyone can compete.
What's more, with a number space between 38 and 45 bits (depending upon
what eventually happens) lots of people would have lots of opportunity
to get very rich, before anyone even notices that there are any duplications

  | As Bob's draft says, safeguards are needed to ensure that if it makes
  | a profit, it would be used for the public good.

It has to make a (BIG) profit - as the price has to be set artificially
high to prevent speculators simply claiming the entire number space, and
then reselling it all later, once it is exhausted.   Even with a 45 bit
number space, that's entirely possible, if the numbers were sold at
anything even approaching the real cost of doing so.

This is exactly what will encourage others to shove in their oar.

  | Well, I think we also have running code to prove that people value
  | hanging off the unique root, since despite dire predictions we still
  | seem to have a single namespace.

What running code?   All the obvious examples are where the uniqueness
is useful for something.

  | What you get for 10 Euros is *only* the assurance that the same
  | number hasn't been given to anyone else.

But how could anyone offer that assurance, when someone else could be
giving out the same number?   Someone with just as much right as the
IANA to allocate meaningless numbers...

The only assurance that anyone could really offer is that they won't
allocate the same number to anyone else.    Is that worth the 10 Euros
when someone else will offer the same thing for 5 ?

  | Another way to look at it is (using my example prefix) that IANA
  | has delegated FC00::/8 to a particular registry, and that
  | registry has chosen to sell non-aggregatable /48 prefixes
  | for 10 Euros each. If they stay in business, fine; if they go
  | bust, that's fine too.

But why does Mr X care what IANA has allocated, he can sell /48's in
FC00::/8 just as easily as the "particular registry" can, and most
likely can sell them a lot cheaper, as he doesn't care if speculators
buy out his whole supply (in fact, X would probably love that, and
even offer a discount, at even 0.01 Euro's per number, 2^40 numbers
adds up to a tidy sum!)

As I said in an earlier message, if we want to pretend that the numbers
are going to be unique, we have to have something that only works if
they are in fact unique.   Here there's nothing.


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 majordomo@sunroof.eng.sun.com