[nycbug-talk] IPv6 Followup

Miles Nordin carton at Ivy.NET
Mon Nov 12 11:46:12 EST 2007


>>>>> "ys" == Yusuke Shinyama <yusuke at cs.nyu.edu> writes:
>>>>> "sr" == Scott Robbins <scottro at nyc.rr.com> writes:

    sr> native IPv6 to home users, but they sort of kill the whole
    sr> point of the thing by giving you a /128. (Yes, one single IPv6
    sr> address. And dynamically assigned, no less.)

It's a lot of extra work, complexity, and user-support to set up
something like that with IPv6---custom infrastructure, built for the
``experiment,'' just to assign in this strange way.  The registry's
address handout rules give no incentive for assigning longer than /56.
I can't think of any reason NTT would work so hard, unless they
intentionally want to stop home users from having reachable global
addresses.  Maybe there's some other reason I can't imagine.  but, it
is really a lot of work for no obvious benefit to them.

    ys> no one can ever come up with a good use of v6 that has enough
    ys> appeal for general public (other than hacking).

In view of the above, I'd say you have it backwards.  I think the
imagined and unimagined usefulness of it to home users is exactly the
reason it's _not_ happening.  It's threatening to the biggest ISP's,
though I don't know why.  Imagined reasons:

 * The one we care about: they sell static addresses and sell port
   unblocking for huge per-profit fees, like double what Interweb
   users pay.  They don't want a second-class v6 hosting option to
   exist.

 * They want a clear distinction between hosting and browsing, so the
   browser ISP Level 3 can charge the hosting ISP Cogent for their
   users?  

 * They don't want filesharing to work without setting up complicated
   port-mapping mumbo jumbo?  (probably more the cable company than
   the phone company though)

 * They're afraid of a plug-in ``hosting appliance?''  

 * They want to mediate your communication by making sure everyone
   meets each other on some web ``property'', youtube, itunes,
   myspace, iDisk.mac.com, gmail---no direct communication between
   houses?  

 * They have long-term plans to move your data off your computers and
   onto theirs, and then rent it back to you?  

 * They're blindly afraid of applications they can't imagine
   threatening their own web ``properties''?

   ``The future is like walking down some unlit corridor, and it gets
    darker and darker as you move into it.''
          -- Jack Valenti.  

some guesses, but I don't know why.  /128 reaks of fear, though.

Next, an earlier slide said the plan for the last stages of the v4
crunch is a cap-and-trade system like carbon emissions.  Your
number-rental fees will go up and down each month like the price of a
kilowatthour.  There could be speculation, international trade, a
heavily-regulated market full of complicated tricks for sophisticated
players.  I think Alex's ``important'' people find this sort of thing
infinitely more exciting than IPv6.

    ys> "IPv6 all connects the electric stuff at your home together!
    ys> TVs, ACs, and lamps..."

in reality it connects together all the many ``boxes'' the cable
company gives you as they slowly take over everything in your home.
phones, stereo, three televisions, all your computers.  There aren't
enough rfc1918 addresses, and they want to retain central control of
all your appliances even after you buy them.  It complicates things
for them to reuse addresses between homes.  The devices might not even
be on the Internet, but their own corporate/customer network is so
big---as single companies buy up the infrastructure of entire nations
and continents, the real technical need is for a convenient way of
assigning bigger blocks of numbers than a /8 to a single entity.

The celfone companies need it, want it, are using it already?, for the
same reason.  Isn't it part of WCDMA?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 304 bytes
Desc: not available
URL: <http://lists.nycbug.org/pipermail/talk/attachments/20071112/18f48e8c/attachment.bin>


More information about the talk mailing list