Measuring World IPv6 Launch Participants From All Over The World
Last year, we measured selected participants from World IPv6 Day on 8 June 2011. For this years World IPv6 Launch event on 6 June 2012 we're going to do the same, and make the results available on http://ipv6launch.ripe.net . As the World IPv6 Launch has a broader focus then the event last year, we're focusing these measurements on the websites that will be (or already are) accessible over both the legacy IPv4, and the newer IPv6 network.
For this year's event we picked participants from the official participants list, so to have geographic diversity and diversity between those that are already 'doing it' and those that are not yet.
We're already seeing people going ahead and publishing AAAA records for their websites, which enables IPv6 clients to make connections to the IPv6 webservers. This is visible in Figure 1 (click on it for a larger version), where blue represents no AAAA records (i.e. the service is only accessed over IPv4), and the yellowish green representing AAAA records visible from all our vantage points. Other blends of blue and green are shown when some but not all of our vantage points see AAAA records, and white is used when we don't have measurement data. Sites that we see already made the switch are Facebook (21 May), Netflix (partially on 22 May, D-Link (24 May), Swisscom (25 May) and Huawei (28 May). So one could say World IPv6 Launch is already well underway!
Figure 1: Is a AAAA record announced in DNS for selected World IPv6 Launch participants. Blue means no AAAA records are seen by any vantage point, yellowish green means all vantage points saw AAAA records for a participant.
When looking at availability of information over HTTP in IPv6 to the participants, we see some issues still remaining, but overall things are looking pretty good. Figure 2 is a snapshot (2012-05-29 12:00 UTC) of the results of HTTP fetches over IPv6. Rows represent destinations (i.e. web sites), columns are vantage points we measure from. Colours represent success (green), failure (red), dns-problem (orange), no IPv6 address in DNS (blue), not measured (white). While some of the failures that we measure are transient, we do see a few rows (web sites) and columns (vantage points) with a lot of 'red' cells, indicating failures. Some of these we can explain and are working with the operators at these locations to fix. For instance one vantage point isn't getting anywhere in IPv6, and traceroute6 stops in their upstream network, which is a pretty strong indication that something needs to happen in this upstream network to get this working again.
Figure 2: Snapshot of HTTP measurements over IPv6 to websites of selected World IPv6 Launch participants. Rows represent destinations (i.e. web sites), columns are vantage points we measure from. Colours represent success (green), failure (red), dns-problem (orange), no IPv6 address in DNS (blue), not measured (white).
If you have any feedback on these measurements, or suggestions for improvements, please comment below. Please also see other activities the RIPE NCC is involved in for World IPv6 on http://www.ripe.net/ipv6launch .