|
We are running out of IPv4 Internet addresses, but who is using them all up? The IP address space usage per capita differs greatly between nations, which points to a digital divide.
If we would distribute IPv4 addresses uniformly over the world population, there would be less than 1 address per person. In fact, on the average, 0.54 addresses would be available per person (including all babies, etc). In 2006 this number was 0.57, population has increased in the meantime. The actual use is 0.45 (up from 0.32 in 2006). For a more detailed analysis of this history, see my 2006 analysis.
But addresses are not uniformly distributed. There is a digital divide between western countries and developing countries. The top user (not counting some mini states) is the US, where wasteful pre-CIDR address allocations have led to an average of more than 4 addresses per person. A few dozen countries have more than their fair share (i.e. 0.54) addresses per person, and almost two dozen have more that 1 address per person allocated.
Utilization is still growing rapidly. Over 30, mostly developing, countries have more than doubled their address utilization per capita in the past 4 years. In addition, most developed countries (with the notable exception of the US) have seriously (30-60%) increased their utilizations.
This growth is clearly unsustainable within the IPv4 address space. Not every country can have these utilization levels. The hunger for new addresses is greatest in China (currently at 1 IPv4 address per 4 inhabitants) and India (1 address per 53 inhabitants). To put these at the modest level of 1 address per inhabitant requires more than 2.2 billion addresses, where there are currently only 290 million left, according to http://www.potaroo.net/tools/ipv4/. Given these numbers and the overall strong growth, any hopes of being able to reuse space that is allocated but not used (i.e. pre-CIDR) are futile. This demand dwarfs the entire US allocation.
IPv6 is supposed to solve the address space problem. Is it progressing? For IPv6 allocations the per capita metric is a bit moot as there is no scarcity. Still, there is an interesting divide. Quite a number of countries have no allocated IPv6 address space at all. A few have pilots, with utilizations less than 1 percent. The largest utilizations are in Sweden and in the Asia Pacific region, where a number of countries have hundreds of IPv6 addresses per inhabitant. The US is in 8th place, after Vietnam and Indonesia but, somewhat surprisingly, before China.
The full data of this analysis can be found in a Google Docs spreadsheet. I’d also like to acknowledge Iljitsch van Beijnum who has provided the address allocation data.
Sponsored byCSC
Sponsored byVerisign
Sponsored byDNIB.com
Sponsored byIPv4.Global
Sponsored byWhoisXML API
Sponsored byVerisign
Sponsored byRadix
Hi,
I fully agree with you IPv6 cannot solve the address shortage problem Ipv4and 6 both have inherent shortcomings. I am working in this area via a net service which will have location details and allow multiple users to have the same username by providing the user with a relatively small and remember able number, but the number analogy used will be absolutely in -exhaustible i.e. no digits will have to be added to it ever like telephone numbers. So cloud computing can be universal (though terms used today are private and public) Vint Serf is talking about common standards but I have my doubts about getting people to work together, with their revenue model and technologies in stealth mode.
This can only be possible if there is a common platform.
Virendra
Do review previous new IP addressing schemes by, for example, Jim Fleming first. Thank you.
It’s not clear to me that an analysis of national per-capta utilization of IPv6 address space has any value whatsoever. For one thing, addresses are allocated on a network provider basis as network providers determine they have need for IPv6 addresses, not on a country basis based on population, economic development, geography, or other criteria. The fact that some countries do not have any IPv6 addresses attributed to them merely means the network providers in those countries have not yet determined they need IPv6 addresses. As soon as they feel they need IPv6 addresses, they can go to their regional Internet registry and obtain a minimum of 79,228,162,514,264,337,593,543,950,336 addresses if they are an ISP or 1,208,925,819,614,629,174,706,176 if they aren’t.
Secondly, the attribution of address to country is, at best, an educated guess and can tend to be quite misleading. The country designated in address block allocations is typically based on the headquarters of the requesting organization and, particularly in the case of multi-national ISPs, does not necessarily correspond to where the addresses are being used. This is why some folks discover they can’t gain access to sites like Hulu even though they are in the ‘correct’ country.
Finally, the amount of IPv6 is so vast that given any rational allocation model, scarcity of IPv6 addresses is simply not an issue the Internet community should ever face. To try to give you an idea of the sheer scale of IPv6, even if the current address consumption rate of IPv4 were to be multiplied by 1000, the 1/8th of the IPv6 address space that has been designated for “global unicast” use would last over 100 years. That leaves the remaining 7/8ths of the IPv6 address space. We’ll run into other scaling problems long before we get anywhere near running out of IPv6 address space.
(speaking personally and representing no one but myself)
You can be a network in a developing economy in asia or south america, or one in the USA or Western Europe. You still follow roughly similar procedures to apply for and acquire any sort of address space, IPv6 fine - or even IPv4 if you want that.
The concept of a “digital divide” in IP allocation is a hackneyed fallacy, and implies that developing economy networks are denied any IP space.
And as you can probably tell from the statistics Iljitsch van Beijnum provided you, the old shibboleth about MIT having more IPs than China ceased to be true several years back.
Gents, I wrote a reply several days ago, but it hasn't appeared, so I have written a separate blog post that should be visible on CircleID shortly.
“hundreds of IPv6 addresses per capita” should read “hundreds of /64 IPv6 blocks per capita”.
The number of adresses in a /64 is equal to the number of addresses in the IPv4 space, squared.
Which volumes like these, measuring IPv6 allocation growth is at best a bit indicative. Yet, to the extent that registries follow similar policies, these numbers are some indication of IPv6 growth.