So today (June 8th, 2011), is the first “World IPv6 Day”. This is (was) a day where the techies all over the world were encouraged to try and see what worked with IPv6 and what didn’t. The good news is that the publicity has definitely helped demonstrate that this is an important topic for the future of the Internet. As more and more non-techies are aware, we ran out of IPv4 addresses a while ago (yes yes, I know that is not entirely an accurate statement), and the only long-distance solution is something called IPv6. Which I won’t explain here.

So what did I do to celebrate this monumental event? First, I checked the “who’s supporting it” list and was happily surprised to see a Fairly Long List of participating organizations. Granted, compared to the much longer list of organizations connected to the internet, this is nothing, but it’s a start… Then, I launched my web browser and hit a few random sites to see how they worked over my IPv6 connection. I was happy to see they worked fantastically.

But, I thought, what a boring test. How do I know that my dual-stack IPv4/IPv6 machine is really doing everything it can to reach these sites using only IPv6? I had two choices: pull the IPv4 plug out (no, non-techies, it’s not actually a different plug; sorry for the confusion) or do the right thing and actually test the real data. So I did the right thing. Or at least part of the right thing.

I quickly hacked up a script (now available from my GitHub ipv6day repository) to test a few important elements of how well an organization would truly fair in an IPv6 only world. Connectivity can fail to anyone at any time, and I wasn’t trying to test connectivity. I wanted to test whether they were truly advertising all the needed services of their organization as IPv6 capable. So I tested 3 things (arguably the 3 most important things):

• Did they have a “www” record with a AAAA address (which is an IPv6 address)?
• Did they have at least 1 NS (DNS) entry for their domain with a AAAA record?
• Did they have at least 1 MX (email) entry for their domain that was reachable by IPv6?

Turns out, most sites concentrated on only the first question and skipped the other two. Not entirely a true commitment to testing IPv6. Still, I suppose, better than nothing. But I still won’t list them as “success cases”. If you had only IPv6 on your machine, you certainly couldn’t read their website without at least the first two, and you couldn’t send them mail complaining about it unless you had the last two.

As long as I was at it: DNSSEC

I figured as long as I was testing things, I wanted to test out my ability to use my other new favorite technology: DNSSEC. How many of these domains would at least have:

• At least one published DNSKEY?
• A parent that had a DS record pointing to them?

Note that I wasn’t testing the actual data. Just “if they were thinking about it”, as I wasn’t even checking to see if DNSSEC signatures were being published. And it’s not entirely fair if their parent won’t accept a DS record to publish (but that’s too bad; fix your parent).

Results

The full result table is long. So what do you do with long results? Summarize them of course! So here’s the summary table (the numbers in green are the number of sites that succeeded in all the tests).

It’d be nice if every one of those entries had maximized the IPv6 tests (3) and DNSSEC tests (2). But as you can see, we fell far short of that.

So who gets the gold stars? Of all the 436 domains that had listed themselves as testable, who actually truly tried their best? It’s only fair that I specifically call out “good job” to those that I consider having passed the “World IPv6 Day Test”. These weren’t hard tests. They weren’t even under undergrad course-level worthy. They’re far below questions that might be given during a 101 class test, and were more like a Elementary School course-level test questions. But those that passed are still worth naming.

41 passed the IPv6 test and 36 passed the DNSSEC test. That’s right, of all the companies that said they’d participate in “World IPv6 Day”, only 41/436 (9.4%) passed my simple tests. i don’t think the techie world scored an a+ today.

But, as in every class, there were a few over-achievers. i’ll call out their names as they cross the stage first, because not only did they pass the 3 simple ipv6 tests, they also passed my dnssec tests, which was really a pop-quiz they didn’t know they’d be taking. Kudos to the folks on this list! I passed out the most gold stars to this fine list of students:

Domain	IPv6	DNSSEC	Both
gotanet.se
imperial.ac.uk
lansstyrelsen.se
leissner.se
lst.se

And now the list of companies that at least passed either the full set
of IPv6 tests or the full set of DNSSEC tests:

Domain	IPv6	DNSSEC	Both
afilias.info
antixlabs.com
beamartyr.net
bluecatnetworks.com
broeltal.de
caib.nasa.gov
census.gov
cesca.cat
commerce.gov
cysols.com
dhs.gov
doi.gov
dol.gov
dpf.gov.br
ed.gov
energy.gov
faa.gov
fab-corp.com
feb.gov
gigatux.com
gsa.gov
gustavus.edu
hostingxs.nl
hostmaster.ua
httrack.com
hurricanelabs.com
ig.com.br
iltalehti.fi
indiana.edu
info.info
infoblox.com
inl.gov
internet2.edu
ipv6world.nl
isoc.se
jobs.qualcomm.com
jp.apan.net
kernel.org
km.nasa.gov
lemonentry.eu
lmu.de
lsu.edu
midlandcomputers.com
neustar.biz
nist.gov
nlm.nih.gov
opm.gov
orion.on.ca
pir.org
plurk.com
proofpoint.com
reading.ac.uk
rohitab.com
sanger.ac.uk
skymarket.co.uk
state.gov
supranet.net
t-online.de
tamagawa.jp
thehavennet.org.uk
tunix.nl
ucf.org
ulak.net.tr
ultradns.com
verisigninc.com
w3.org
webmetrics.com

Detailed Results

And for those that want to look at all the individual records, I present the full test result table: