Telcos had one circuit, internet increased bandwidth and made things
more efficient, what are the differences?

When we run out of IPv4 addresses will the Internet stop working? 

The purpose of NAT is to increase the number of effective IP
addresses, but it's hard to manage, can't reach them. Convoluted.

Rode on a plane, couldn't connect to the Internet, NAT was at fault.

Marketplace for address blocks. Are we allowed to sell them? Might be
illegal, but won' stop the market?


Should we sell them, even if we're allowed to? End users and
developing nations will be left out.

Market for IPv4 addresses won't fix the running out issue.

NAT works for client-initiated request.  Outside coming in, without
the client-initiated request, is problematic.

NAT is like the old phone systems where people shared the same number (party-line).

Engineers don't like NAT, even though it works for most users.

Co-existence Model: both ipv4 and ipv6 can work together.

Some networks use IPv6 (cable in Japan, but not many others).

Policy makers and governments understand.

Why isn't IPv6 used? Vista failed, might have helped. Chicken/egg, why
should I go and do it, it will cost, no one else is doing it.

Narten says: little can be done to force deployment, reality: business
and economic models drive things, where's the ROI to deploy/adopt
IPv6.

--
How to introduce new technologies?

Need transition plan, how will we do global/internet scale adoption
(e.g., of IPv6) given cost?

Critical components upgraded first: must give these component-owners reason
to upggrade.

Browser was great for Internet because...easy to deploy, security built in, e.g., TLS

Why not upgrade? NOT internet problem, ROI, change is hard, need transition strategy.

We need to show order of magnitude improvement: performance, profit, etc.

=========
Tragedy of the Commons: Challege of Route Scaling

(defn of trag of commons) destroy shared, limited resource

DFZ in internet core --- explicit entries for all prefixes.

If we use Provider Aggregate Addressing: then we wouldn't need to many
addresses.  But, there are roughly 300,000 prefixes currently. Verizon
knows how to deal with Verizon, pass the buck for everyone else. How
do we know where to go? Routing table scalability.  

So, what is the DFZ? Duke wants to be independent from Verizon, not dependent.

Big routers are expensive and don't conform to moore's law.

BIG ISPs are worried that router tables are getting too big.  We won't
be able to reach everyone.  

Worry: Internet is the commons, tragedy will ensue soon?


ICANN has only loose oversight over RIRs, the RIRs (five) hand otu addresses, not
regulated by governments, no follow-up to ensure RIRs follow up on allocated
addresses.

Policy discussions: I want liberal PI space, (provider independent)
but ISPs won't let it happen.

ICANN coordinates parameter registries (what kind of packet we have).

ICANN: ensure uniqueness, stability, and security of identifier system.

-- Why ICANN? Govt get it out. Experiment in private sector
governance. Don't send it to ITU (bureaucrats). Keep ICANN in hands of
technology/engineerin people.

alternatives to ICANN? Nothing seen that would be better. Brazil,
China, grandmas, everyone -- it all works the same.

We should not break DNS: single root. End Users want a single global name space.
We want it the same everywhere.

We should have 'alternate roots'. People understand the value of a
single, consistent global namespace. Any claimed benefits ignore hard
policy (and techincal) questions that ICANN deals with.


Coordinating the global numbering system: ICANN is pretty open, Policy
Development PRocess (PDP) is institutionalized within ICANN.

ICANN cannot turn off the internet, no influence over ISPs

ICANN doesn't control root servers.

Most of the internet is run by business, you can't tell them what to do.

Ultimately the US Government has oversight on what ICANN does. So, USG
could take .ru out of the root zone. But, what would happen? Lots:
UN-level crisis, world backlash, alternate roots by Russia, others
would follow Russia, current root operators might ignore the USG.


In practice concern over USG control is overblown.

  but, political discussions aren't rational.

MOU ICANN and USG expires on Sept 30. Is this a problem? Will someone
in congress give away the internet? No.

Challenges for ICANN: need to increase legitimacy.
What can we do to improve it?

problems:

IPv6
DDOS
Route Scaling
Misguided Government regulation
Robustness and Security of Internet infrastructure
  What about net neutrality? Don't disadvantage skype if you're verizon
  What about SPAM?
  Should p2p networks mess up my email?
  	 QOS ok.