►
We used up 105.57 million addresses in the first half of 2010. So we're on track to beat last year's 203.4 million by a few percent this year. If nothing changes.
Full article / permalink - posted 2010-07-02
▼
On january first this year, the number of available IPv4 addresses was 722.18 million (see the 2009 IPv4 address report). On july first, this was 616.61 million, made up of 268.44 million addresses in the 16 /8 blocks left in the IANA global pool and 348.17 million given out by IANA to the five Regional Internet Registries that the RIRs haven't given out to end-users yet. So we used up 105.57 million addresses in the first half of 2010. That means we're on track to beat last year's 203.4 million by a few percent this year. If nothing changes.
A problem with determining these numbers is that ARIN "backdates" their address allocations sometimes: when they give out a /16 in 2008 and then another /16 in 2010, this will often show up as a single /15 block with a 2008 allocation date. So just listing all the allocations for a given year would always show a smaller number than the actual decline of the free address pool. Strangely, the opposite is happening currently: the total number of addresses with a jan - jun 2010 allocation date is 108 million, 2.43 million more than the decline of the free address pool. This suggests that at least 2.43 million addresses given out in previous years have been returned in the first half of 2010. This number isn't easily derived from the published numbers, so I'm not going to look into it now. However, I'll look at the returned address space in some detail in the future.
Also interesting: last year, LACNIC gave out 10.98 million addresses and AfriNIC 5.99 million. But halfway through 2010, they're already pretty close to last year's totals: 10.38 and 4.67 million, respectively. China got 50.66 million addresses last year, but is now at 18.96, while Korea got 10.95 million last year and has already bested that at 11.68 million for the first six months of the year.
Permalink - posted 2010-07-02
A story I wrote for Ars Technica about depeering between Cogent and Telia, with some info about BGP hot potato and cold potato routing.
Permalink - posted 2008-03-21
►
In january, Geoff Huston wrote to the NANOG list:
George Michaelson, Randy Bush and myself have successfully tested the
implementation of 4Byte AS BGP on a public Internet transit. The
above BGP RIB snapshot was taken at a 4Byte BGP speaker in North
America, showing a transit path across AS 1221, AS 4637, AS 1239 and
AS 3130 , with correct reconstruction of the originating AS at the
other (4Byte AS) end.
At the time of this writing, their prefix is no longer visible in the global BGP table...
Full article / permalink - posted 2007-04-12
▼
In january, Geoff Huston wrote to the NANOG list:
George Michaelson, Randy Bush and myself have successfully tested the
implementation of 4Byte AS BGP on a public Internet transit. The
above BGP RIB snapshot was taken at a 4Byte BGP speaker in North
America, showing a transit path across AS 1221, AS 4637, AS 1239 and
AS 3130 , with correct reconstruction of the originating AS at the
other (4Byte AS) end.
At the time of this writing, their prefix is no longer visible in the global BGP table, but
telnet to route-views.oregon-ix.net and type:
show ip bgp regexp _23456_
Until the Route Views server is upgraded to support 32-bit ASes, this will give you
all the prefixes/AS paths with a 32-bit AS number in them. Currently, those are:
- 84.205.88.0/24, "RIPE-NCC-RIS 4-byte ASN testing prefix", advertised from AS 196615 aka AS 3.7
- 145.125.0.0/20, "SURFNET-TEST-NETWORK 4BYTE ASN", advertised from AS 196613 aka AS 3.5
There are now patches to make OpenBGPD 32-bit AS capable and also patches for Quagga.
For more information on 32-bit ASes, have a look at an earlier article on BGPexpert.com.
Permalink - posted 2007-04-12
►
Old dogs can learn new tricks. That's a good thing, because securing inter-domain routing requires a whole bag of them. After lots of talk about S-BGP and soBGP over the past years, more recently, work in the IETF on making inter-domain routing more secure has shifted to a different approach.
Full article / permalink - posted 2007-03-21
▼
Warning: spoiler. Last week's episode (#74) of MythBusters showed that, in fact, old dogs can learn new tricks. That's a good thing, because securing inter-domain routing requires a whole bag of them. After lots of talk about S-BGP and soBGP over the past years, more recently, work in the IETF on making inter-domain routing more secure has shifted to a different approach. The relatively new secure inter-domain routing (sidr) working group is now working on providing a public key infrastructure that makes it possible to link an IP prefix to an origin AS with certificates. In the future, this mechanism may be used in S-BGP, soBGP or a similar mechanism, but in the mean time, it allows generating and validating filters. Although it's possible to choose arbitrary trust anchors, the idea is that IANA and the RIRs will serve as certificate authorities as they are the ones giving out the address space and AS numbers. Although the basic idea is simple enough, I'm slightly worried about how this is going to work in practice, because the underlying mechanisms are very complex, and not something "BGP people" are likely to be familiar with. Have a look at the sidr page on the IETF website and the links to the current drafts to get an idea. A good one to start with is draft-ietf-sidr-arch-00.txt or "An Infrastructure to Support Secure Internet Routing". (Link to the latest version.)
Permalink - posted 2007-03-21
OpenBSD, the security conscious sibling in the BSD operating system family, has its own BGP daemon implementation: OpenBGPD.
Permalink - posted 2007-01-30
As Zebra progress has been glacial, a group of people created a fork under the name Quagga. Quagga is more community-based and a somewhat better choice than Zebra in an operational environment.
Permalink - posted 2007-01-30
older posts
- newer posts
