I want to thank you for letting the EOS Networks Prototyping lab (ENPL) sponsor mirror sites of the Astronomy Picture of the Day (APOD) and Earth Observatory (EO) web pages for World IPv6 Day on June 8, 2011. Going into this we had no idea what would happen. The outcome could have ranged from a mere trickle of interest to normal levels of IPv6 activity to large-scale denial-of-service attacks. Fortunately, we had no network or system failures and we were able to provide service during the entire 24 hour period. Those of us who provided NASA web sites for the event were told to support them in dual-stack mode, that is, to make content available using IPv6 as well as IPv4. We logged every access attempt to the ENPL World IPv6 Day webserver including each user’s IPv4 or IPv6 address. We used Google Analytics to provide additional feedback of usage and the distribution of page visits across the two mirror sites. A quick summary of the results follows:
Total user connections to the ENPL webserver: 59,667
Total IPv6 user connections to the ENPL websever: 34,501
Total IPv4 user connections to the ENPL webserver: 25,166
Total unique IPv6 source addresses: 234
Total unique IPv4 source addresses: 1690
Three IPv6 users accounted for 19.6% of the total IPv6 access attempts.
Two IPv4 users accounted for 18.25% of the total IPv4 access attempts.
The Google Analytics results fall way short of the counts from the ENPL traffic logs. However, this discrepancy can be easily accounted for. Google Analytics requires javascript in order to do accounting. If a browser did not have javascript enabled then the the access would not be counted by Google but even more revealing is that if the user merely ran a script that connected to port 80 successfully, via telent for example, then Google Analytics didn’t count them either. Given the distribution of hits across the total number of user addresses for each protocol, it’s readily apparent that users had set up scripts to hit the World IPv6 web sites. I have an overwhelming hunch that the event sponsor, The Internet Society (
http://www.isoc.org), did just that. See
http://www.worldipv6day.org/participant ... index.html for more information.
From a network perspective, and World IPv6 Day was very much about globaI IPv6 reachability , it didn’t matter that much if the source of the IPv6 connection attempt was automated or mouse-driven. The internet was able to operate quite easily in dual stack mode with 99% of the World IPv6 Day participants web sites reachable using IPv4, 92% reachable using IPv6, and 94% had resolvable AAAA DNS records (the IPv6 name-to-address mapping record). This is very good news for the internet as the migration to IPv6 continues.
From the perspective of counting mouse-clicking users behind their browsers, the ENPL results aren’t so awe inspiring. Yet there is useful data to be gleaned about IPv6 availability to the end user, if the user is IPv6-enabled, and if the required IPv6 network services are available for the user to make seamless end-to-end IPv6 connections. There remains quite a bit of Google data that needs to be analyzed and it’s more complex than simple counts and percentages so I can only provide a very quick summary of the Google Analytics results (taken with a grain of salt as the Google data is somewhat confusing):
Total Visits
APOD – 1899
EO – 326
Total Protocol Events
APOD – 6434
EO – 84
Without more analysis It’s not clear to me how many visits were from IPv6-only clients, IPv4-only clients, or a dual-stacked clients.
Over 70 countries had at least one successful access attempt. About 53% came from the US, 7% from Canada, Mexico and South America, 30% from Europe, 5% from Asia, and the remaining 5% from Oceana, Africa and places unknown.
We appreciate your interest and support of this global event and we will be publishing a more detailed report of the event in the near future.
Thanks,
George
-------------------------
George Uhl
ESDSIS Network Engineer
Code 423
NASA Goddard Space Flight Center
Greenbelt, MD 20771
Phone: 301-614-5155
Fax: 301-614-5700
email:
george.d.uhl@nasa.gov
-------------------------
I want to thank you for letting the EOS Networks Prototyping lab (ENPL) sponsor mirror sites of the Astronomy Picture of the Day (APOD) and Earth Observatory (EO) web pages for World IPv6 Day on June 8, 2011. Going into this we had no idea what would happen. The outcome could have ranged from a mere trickle of interest to normal levels of IPv6 activity to large-scale denial-of-service attacks. Fortunately, we had no network or system failures and we were able to provide service during the entire 24 hour period. Those of us who provided NASA web sites for the event were told to support them in dual-stack mode, that is, to make content available using IPv6 as well as IPv4. We logged every access attempt to the ENPL World IPv6 Day webserver including each user’s IPv4 or IPv6 address. We used Google Analytics to provide additional feedback of usage and the distribution of page visits across the two mirror sites. A quick summary of the results follows:
Total user connections to the ENPL webserver: 59,667
Total IPv6 user connections to the ENPL websever: 34,501
Total IPv4 user connections to the ENPL webserver: 25,166
Total unique IPv6 source addresses: 234
Total unique IPv4 source addresses: 1690
Three IPv6 users accounted for 19.6% of the total IPv6 access attempts.
Two IPv4 users accounted for 18.25% of the total IPv4 access attempts.
The Google Analytics results fall way short of the counts from the ENPL traffic logs. However, this discrepancy can be easily accounted for. Google Analytics requires javascript in order to do accounting. If a browser did not have javascript enabled then the the access would not be counted by Google but even more revealing is that if the user merely ran a script that connected to port 80 successfully, via telent for example, then Google Analytics didn’t count them either. Given the distribution of hits across the total number of user addresses for each protocol, it’s readily apparent that users had set up scripts to hit the World IPv6 web sites. I have an overwhelming hunch that the event sponsor, The Internet Society (http://www.isoc.org), did just that. See http://www.worldipv6day.org/participants-dashboard/index.html for more information.
From a network perspective, and World IPv6 Day was very much about globaI IPv6 reachability , it didn’t matter that much if the source of the IPv6 connection attempt was automated or mouse-driven. The internet was able to operate quite easily in dual stack mode with 99% of the World IPv6 Day participants web sites reachable using IPv4, 92% reachable using IPv6, and 94% had resolvable AAAA DNS records (the IPv6 name-to-address mapping record). This is very good news for the internet as the migration to IPv6 continues.
From the perspective of counting mouse-clicking users behind their browsers, the ENPL results aren’t so awe inspiring. Yet there is useful data to be gleaned about IPv6 availability to the end user, if the user is IPv6-enabled, and if the required IPv6 network services are available for the user to make seamless end-to-end IPv6 connections. There remains quite a bit of Google data that needs to be analyzed and it’s more complex than simple counts and percentages so I can only provide a very quick summary of the Google Analytics results (taken with a grain of salt as the Google data is somewhat confusing):
Total Visits
APOD – 1899
EO – 326
Total Protocol Events
APOD – 6434
EO – 84
Without more analysis It’s not clear to me how many visits were from IPv6-only clients, IPv4-only clients, or a dual-stacked clients.
Over 70 countries had at least one successful access attempt. About 53% came from the US, 7% from Canada, Mexico and South America, 30% from Europe, 5% from Asia, and the remaining 5% from Oceana, Africa and places unknown.
We appreciate your interest and support of this global event and we will be publishing a more detailed report of the event in the near future.
Thanks,
George
-------------------------
George Uhl
ESDSIS Network Engineer
Code 423
NASA Goddard Space Flight Center
Greenbelt, MD 20771
Phone: 301-614-5155
Fax: 301-614-5700
email: george.d.uhl@nasa.gov
-------------------------