IPv6 momentum: a story of pool and poules[*]

This is part of my series on [http://v3.tronche.com/section/ipv6/foreseeing-ipv6-deployment (Trying to) foresee IPv6 deployment]. Can we identify populations of IPv6 users such as IPv6 adoption logically follows from one population to the next ? [*]poule = the french for chicken, and IPv6 is a chicken-and-egg problem, like the introduction of many new technologies where you need several components for the cocktail to ignite. IPv6 has no business case, except business continuity. There are additional costs, and no additional revenues. Stated that way, it's no wonder why IPv6 has yet to take-off. Roughly speaking in the case of IPv6, the components are: the service providers (like gmail, youtube and consors)[1], and the ISPs. The ISPs have no incentive to enable IPv6 for their customers, since this will give the customers no additional services in the first place, and the service providers have no incentive to enable their services over IPv6, since no one is connected through an ISP with IPv6. How do you trigger a business case in that situation ? Well, you look for niches that are ready to embrace the change, and you jump from a niche to another one in [http://en.wikipedia.org/wiki/Everett_Rogers Rogers] / [http://en.wikipedia.org/wiki/Crossing_the_Chasm Moore] style. In the search for a taxonomy of IP address users relevant to IPv6 adoption, I propose to refine the service provider / ISP distinction based on the size and usage of the address pools done by the various entities. Let us first notice that regions are always more or less relevant, because the big address pools (the [http://en.wikipedia.org/wiki/Regional_Internet_registry RIR]s) are regional. Now, the entities: * (fixed-line) ISPs: they need one IP per customer, typically have millions of them, and allocate addresses in large chunks. Their business is mostly mature, in term of market saturation, in the developed world, if such a thing exists. But it is ''not'' saturated in countries like China. Although the size of the chunks they allocate mean they can't get addresses anymore in many parts of the world (not only Asia), most of the time, they're sitting on a multi-years stock. * Big service providers: Google, Facebook, you name it. They have a world-wide presence, they need a bunch of addresses, but certainly they have enough to hold for several years. * Hosters: they rent servers to users. In some sense, they are like big service providers, but they have less flexibility. First they may not have a world-wide presence, so they may be stuck with their RIR for address allocation. Secondly, big service providers have opportunities to play with their stock of addresses, trying to scratch a few at some places to put them where they are needed. Hosters in contrast have to provide one or several IP address for every dedicated server they sell. * Cell phone operators. They do have a problem. Mobile data has taken off a few years ago with the advent of smartphones, machines able to do something useful with a mobile data connection, where certainly the iPhone led the way. And there are more mobile terminals that fixed lines. For now, cell phone operators have massively relied on NAT, so IPv4 depletion hasn't been a problem for them yet. However, the growing number of always-on terminals put NAT under pressure, and LTE / 4G will kill it. '''I have no doubt that LTE / 4G will start with IPv6'''. There is no sense in putting IPv4 in it right now. * Large companies: they should have enough addresses for their need, but typically depend on their telco / ISP for their employees to access Internet. The main product they use is the VPN ([http://en.wikipedia.org/wiki/Virtual_private_network Virtual Private Network]), which is a whole other story. * Emerging service providers, cloud computing: if you're the next Youtube or the next Google, and you have to deploy thousands of servers now, you have little choice except IPv6. I've already [[ipv6-and-chinese-student-doing-skype-grandpa|argued]] that the first ones to run dry will be ISPs in Asia, and that the fire should then spread to ISPs all other the world. What's the next step ? ---- [1] Sometimes, they are called "content providers". However, I have a hard time calling gmail a content provider. ''I'' am the content provider here, and gmail is a service. So I'll stick to "service provider", even if ISP also means a service provider of a different kind.