This work is not complete yet, and sometimes the graphs show "interesting" artefacts.
If something is wrong, looks weird, needs more work, etc., just let me know.
Right now, this is mostly just presenting the raw curves, and explaining the methodology - no explanations yet on funny bumps in the curves, and no judgement whether something is "good" or "bad".
This is just "the number of prefixes my router has in it's BGP table" - and it's most notably not filtering anything it's receiving from it's peers, so it is likely to have a few more prefixes than a "groomed" upstream would give you (peers are asked to give me "all they receive, without internal prefixes"). Just to point that out: this is not "what AS5539 has in it's normal BGP tables", it's a dedicated monitoring box that takes feeds from external peers.
Taking the RIR "delegated" files to see which region has allocated a given IPv6 prefix, and then mapping BGP prefixes from the routing table snapshots to the individual RIRs. Some prefixes will be mis-mapped because I always use the most recent "delegated" file set, mapping all routing table snapshots over the years against the RIR data for a single day - and if prefixes are returned or move around, mis-mapping of older table entries may happen.
Similar to "by RIR region", these images map the BGP table snapshots to the country attribute given in the RIR "delegated" files. It's most certainly not accurate in all cases (multi-national networks, etc.), but sufficient to get an overview what's going on. The "Top N" countries are selected by "who has the most routing table entries on the last day of the period".
BGP table snapshots are broken down into:
This image is not based on BGP data, it's just plotting the RIR "delegated" files by class (LIR/non-LIR) and RIR. It's background data to understand how much prefixes of which kind are given out in which RIR region, and how the allocation/assignment dynamics are ("how does the curve look like?").
Plotting prefixes that fall into certain "typical" size categories.
Counting unique AS numbers visible in the daily IPv6 BGP snapshots.
An "origin" AS is one that is seen at the end of a BGP path (originating one or more prefixes). A "transit" AS is one that is seen in non-terminal position in a BGP path, providing transit to other ASes. Some do both.
note: the slight spike at 2011/10/09 is most likely not due to something happening to the BGP table, but due to moving the BGP table collection to a new router with new BGP feeds, and thus some ASes got categorized differently.
Cross-checking with the active AS numbers in the IPv4 BGP table.
The end goal is "every participant has IPv6". Right now, the ratio of AS numbers in the IPv6 table to AS numbers in the IPv4 table is more about 11-12%, so there's lots of work remaining.
Potential measurement artifacts: the percentage value assumes that there are not enough IPv6-only ASes in the tables yet to significantly skew the value.
This is just putting the total number of IPv4 prefixes in relation to the number of IPv4 ASes, and vice versa for IPv6.
Some other information sources about IPv6 and BGP:
All BGP table dumps are taken at AS5539 (SpaceNet AG, Munich, Germany), and can be made available upon request.
, SpaceNet AG.