There is a large difference between the number of IP addresses handed out by IANA, and the number of IP addresses actually globally adressable. In the beginning blocks of 16 million addresses were given to any respectable company or organization in the US. Many of these addresses are now used on computers and equipment hidden behind routers and firewalls and could have just as easily been addresses in a private range like 10.xx.xx.xx.

My current internet connection is an GPRS/EDGE connection of a mobile provider with a few million subscribers who funnels all IP traffic through a proxy network with not more than 768 global IPv4 addresses. AOL has used a small number of proxy IPs as long as I can remember. These are examples that it is possible to connect a large amount of clients to the Internet with just a handful of global IP addresses. 4 billion IPv4 addresses is therefore not the same as 4 billion devices connected to the Internet.

Besides the jump to IPv6--which is not an automatic transition--there is also the option of reclaiming some of the assigned but unused IP ranges and my guess is that that will happen in the coming time to extend the lifetime of the IPv4-only Internet as we know it now.

S'true. There's no good reason why my home connection needs to be on a static, externally addressable IP address. Same with the thousands of my neighbors.

Well, there are a lot of good reasons for having public IP addresses on devices. With any NAT / PAT component in between, the so called "end to end principle" is threatened. The "end to end" principle says that the network is extremely dumb, and the end devices are extremely intelligent (contrary to the telephony network, where the network is very intelligent, and the devices are dumb).

A lot of services on the internet depend on that "end to end principle" - for example peer to peer applications. Once there's some NAT in the network, they can't connect directly between each other, and need some "helper" application in the network - which, in turn, makes it more expensive to roll out new services. The "end to end" principle gives full control to the end user.

NATing customers gives a nice level of control over the internet connection to the service provider, and threatens the independence of the user from the service providers network policy.

In a fully NATed network, turning off unwanted applications is a snap for the service provider - all traffic *has* to go through the service providers application gateways. A NATed network is a "gated community".

"predictions". "Experts say". Usual technology journalist waffle. Death of the Internet predicted. Film at 11.

Regards...jmcc

NATing customers gives a nice level of control over the internet connection to the service provider, and threatens the independence of the user from the service providers network policy.

$10 a month for an external IP will let everyone figure out REAL fast how much they need this.

Folks that need it, can pay. Those that don't - I suspect many to most (and myself for certain) don't pay.

Voila, capitalism at it's finest, and a huge swath of IPs released that don't need to be used currently.

$10 a month for an external IP ...

Why not spend some time and money to move to IPv6 for once and be done with this matter for the rest of our lifetime?

A lot of IPs are currently wasted - I recently came across someone who has a separate IP for each site he has ever set up, including several inactive sites. None has an actual need for its own IP.

Another thing a monthly charge would do is help SNI take off, so we do not need a separate IP for each ssl site.

@wheel, it would certainly make the people who have class A blocks think twice.

@iThink, because no one has an incentive to do it until everyone else has.

IBM has more IPs available than the whole of China uses.

It's not just IBM, but a number of corporations which were using Internet in the early stage, like HP, Ford and USPS. Some others like Standford University have already given up their class A address range.

China is actually a good example how a large number of devices can be connected successfully to the IPv4 backbone with only minimal use of globally addressable IPv4 addresses. Their local infrastructure is partly IPv6 which gives a huge amount of local addresses while the connections to the world are still IPv4. Many Russian ISPs are fully NAT based where the customer can get a fixed IP at a premium. Those who came later to the Internet may be better adapted to the near future than those who have been there since the beginning.

Why not spend some time and money to move to IPv6 for once and be done with this matter for the rest of our lifetime?

A fully IPv6 world without IPv4 is an Utopia for the coming 20 years. Too many devices don't have IPv6 networking functionality. It takes one generation before they are all gone. As illustration, there are still Alexa 1000 sites who don't use MX records for their mail servers. MX records were introduced in 1986.

Why not spend some time and money to move to IPv6 for once and be done with this matter for the rest of our lifetime?

because conservation is easier.

I expect ipv6 is going to mean I have to upgrade hardware, and definitely upgrade software. Not interested when the primary reason we need to upgrade is due to wastage.