This archive is retained to ensure existing URLs remain functional. It will not contain any emails sent to this mailing list after July 1, 2024. For all messages, including those sent before and after this date, please visit the new location of the archive at https://mailman.ripe.net/archives/list/[email protected]/
[ipv6-wg] 2010 IPv4 (and IPv6) Address Use Report
- Previous message (by thread): [ipv6-wg] RIPE-501bis
- Next message (by thread): [ipv6-wg] "Requirements For IPv6 in ICT Equipment" comment
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Iljitsch van Beijnum
iljitsch at muada.com
Tue Jan 4 12:30:04 CET 2011
[ (Non-cross)posted to NANOG, PPML, RIPE IPv6 wg, Dutch IPv6 TF. ] On the web: IPv4: http://www.bgpexpert.com/addrspace2010.php IPv6: http://www.bgpexpert.com/addrspace-ipv6-2010.php The IPv4 one is included below: 2010 IPv4 Address Use Report As of January 1, 2011, the number of unused IPv4 addresses is 495.66 million. Exactly a year earlier, the number of available addresses was 721.06 million. So we collectively used up 225.4 million addresses in 2010. 35 of the 256 the /8s that make up the IPv4 address space have the status "reserved". 0 and 127 have special meaning and can't be used for normal purposes. 224 - 239 are used for multicast and 240 - 255 are "reserved for future use". With only about two years worth of IPv4 addresses remaining on the shelves, it would seem that that future is here now, but unfortunately, pretty much all operating systems balk at using a "reserved" address. So unreserving those addresses means upgrading EVERY system connected to the Internet. If we're going to do that, we may as well skip those reserved IPv4 addresses and upgrade to IPv6. Last but not least, there's block 10, which is the largest of the three address blocks set aside for private use. The others, 172.16.0.0/12 and 192.168.0.0/16, don't show up as reserved, but are obviously not available for regular use. This makes the total number of usable IPv4 addresses is (256 - 35) * 2^24 - 2^20 - 2^16 = 3706.65 million addresses. The "IANA global pool" consists of 7 /8s (117.44 million) are still unused (unallocated): 39/8, 102/8, 103/8, 104/8, 106/8, 179/8 and 185/8. But there's also a lot of unused space hiding in the "allocated" and "legacy" categories. Each RIR publishes a list of address blocks further delegated to ISPs or end users every day on their FTP servers. If we add up all those blocks, this comes out to 3210.99 million addresses. So the total number of usable-but-unused IPv4 addresses is 3706.65 - 3210.99 = 495.66 million. Going back to the IANA global pool, these are the changes over the past year: Delegated Blocks +/- 2010 to/status AfriNIC 3 +1 APNIC 42 +8 ARIN 35 +4 LACNIC 8 +2 RIPE NCC 34 +4 LEGACY 92 UNALLOCATED 7 -19 There is an agreement between IANA and the RIRs that each RIR will get one of the last five /8s. APNIC has been getting two /8s every three months like clockwork in 2010. If this continues, they'll be getting numbers 7 and 6 later this month, and then the final distribution will look like this: Delegated Blocks +/- 2010 to/status AfriNIC 4 APNIC 45 ARIN 36 LACNIC 9 RIPE NCC 35 LEGACY 92 UNALLOCATED - At this point, it becomes very interesting what the status of the legacy space is, exactly. The legacy blocks are each "administered" by one of the RIRs, but does that mean that that RIR is free to further delegate that space to ISPs and end users? There are 146.92 million unused addresses in legacy space, including 16.65 million returned by Interop a few months ago. This is the used versus unused address space administered by each RIR: Legacy Allocated total unused total unused AfriNIC 33.55 24.85 50.33 27.06 APNIC 100.66 22.32 704.64 44.38 ARIN 654.31 60.55 587.20 56.21 LACNIC - - 134.22 37.39 RIPE NCC 67.11 5.77 570.43 67.38 IANA 671.09 16.65 - - AfriNIC used up 8.95 million addresses last year. So their current unused allocated space is good for another three years (if nothing changes) and their final /8 is worth another almost two years. If they get to use their legacy space, that buys them another 2.5 years. So unless IPv4 address use <em>really</em> takes off in Africa, AfriNIC will be handing out addresses for at least three or four years. APNIC is at the opposite end of the spectrum, using up no less than 126.22 million new IPv4 addresses last year. Even if they get to use the legacy space they administer on top of three of the last seven /8s and, it's hard to see how APNIC can avoid having to tell people "no" before the year is out. However, there is a caveat: in the 2010 APNIC records, there is 6.65 million addresses worth of space that isn't in the 2011 records. Part of this is address space returned to APNIC. In other cases, an address block delegated in a previous year expands or shrinks retroactively. Depending on what the underlying reason for these changes is, the actual rate at which APNIC and the other RIRs are giving out address space may be different from what it seems to be at first glance. ARIN, LACNIC, and the RIPE NCC used up 54.55, 17.29, and 75.45 million addresses, respectively, in 2010. However, ARIN saw 27.24 million addresses returned, including the 16.65 million from Interop, which is administered in the ARIN records even though the IANA list doesn't reflect this. For AfriNIC, LACNIC and the RIPE NCC the numbers of addresses that came back were 0.31, 0.22, and 22.62 million, respectively. With respect to running out of addresses, it's important to realize that the Pareto principle (the 80/20 rule) applies: out of the 7686 address blocks given out last year, only 392 (5 percent) were blocks larger than 100,000 addresses, but those were responsible for 82 percent of the address <em>space</em> given out. Even when the RIRs are no longer able to give out those large blocks, they may still be able to fulfill the requests for address blocks smaller than 10,000 addresses. Last year, 6425 such blocks were given out, totaling 14.03 million addresses. It really only takes a single address to be in the content business; it's the ISPs that need a continuous supply of new addresses to connect new customers. So the address shortages looming beyond the summer will hit ISPs and their broadband/mobile customers first and foremost, and the content industry to a much lesser degree. The top 15 IPv4 address holding countries: 2011-01-01 2010-01-01 Increase Country 1 - US 1519.53 M 1495.13 M 1.6% United States 2 - CN 277.64 M 232.45 M 19.4% China 3 - JP 186.82 M 177.15 M 5.5% Japan 4 - EU 151.80 M 149.48 M 1.6% Multi-country in Europe 5 (6) KR 103.50 M 77.77 M 33.1% Korea 6 (5) DE 91.61 M 86.51 M 5.9% Germany 7 (9) GB 82.25 M 74.18 M 10.9% United Kingdom 8 - CA 79.53 M 76.96 M 3.3% Canada 9 - FR 79.29 M 75.54 M 5.0% France 10 - AU 49.10 M 39.77 M 23.5% Australia 11 - BR 40.24 M 33.95 M 18.5% Brazil 12 - IT 37.14 M 33.50 M 10.9% Italy 13 - RU 34.66 M 28.47 M 21.7% Russia 14 - TW 31.93 M 27.10 M 17.8% Taiwan 15 (19) IN 28.70 M 19.42 M 47.8% India Because the US holds so much space, the increase of 25 million addresses seems small, but that's still more than 10% of the address space given out in 2010. China's growth is slowing down a little at 45 million addresses last year compared to 50 million in 2009. But other countries in Asia are picking up the slack and then some: Korea keeps using up large amounts of address space, and India is now also picking up the pace. The US now has 47.3% of the address space in use, down from 50.1% a year ago. The other countries in the top 15 collectively hold 39.7%, up from 38%. That leaves 13% for the rest of the world, up from 12%. Note that I slightly changed the way addresses are counted: previously, all the legacy blocks that didn't have an RIR listed were assumed to be used 100%. But with the return of most of the Interop block this is no longer the case: although ARIN isn't listed as administering the 45/8 block, they actually are and only have 45.0.0.0/15 listed as in use.
- Previous message (by thread): [ipv6-wg] RIPE-501bis
- Next message (by thread): [ipv6-wg] "Requirements For IPv6 in ICT Equipment" comment
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
[ ipv6-wg Archives ]