Hi all, at RIPE 65, Lorenzo Colitti from Google raised the question whether Internet users would see significant filtering of AAAA DNS queries or replies, either by their ISP or by their Customer Premise Equipment (CPE). Emile & Philip used RIPE Atlas to provide an answer to this question. https://labs.ripe.net/Members/emileaben/ripe-atlas-a-case-study-of-aaaa-filt... Regards, Vesna
On Thu, Oct 11, 2012 at 7:08 PM, Vesna Manojlovic <BECHA@ripe.net> wrote:
at RIPE 65, Lorenzo Colitti from Google raised the question whether Internet users would see significant filtering of AAAA DNS queries or replies, either by their ISP or by their Customer Premise Equipment (CPE). Emile & Philip used RIPE Atlas to provide an answer to this question.
https://labs.ripe.net/Members/**emileaben/ripe-atlas-a-case-** study-of-aaaa-filtering<https://labs.ripe.net/Members/emileaben/ripe-atlas-a-case-study-of-aaaa-filtering>
Thanks for publishing the results! When you did the AAAA measurements, did you distinguish between the probe not getting a response (e.g., if the dig command timed out) and the probe getting a NODATA response? [I tried to look at the raw data to answer this question, but wasn't able to find out how to download it. I can see the probes that performed the measurement, but not that the results were. Am I doing something wrong?]
On 10/11/12 1:56 PM, Lorenzo Colitti wrote:
On Thu, Oct 11, 2012 at 7:08 PM, Vesna Manojlovic <BECHA@ripe.net <mailto:BECHA@ripe.net>> wrote:
at RIPE 65, Lorenzo Colitti from Google raised the question whether Internet users would see significant filtering of AAAA DNS queries or replies, either by their ISP or by their Customer Premise Equipment (CPE). Emile & Philip used RIPE Atlas to provide an answer to this question.
https://labs.ripe.net/Members/emileaben/ripe-atlas-a-case-study-of-aaaa-filt...
Thanks for publishing the results!
When you did the AAAA measurements, did you distinguish between the probe not getting a response (e.g., if the dig command timed out) and the probe getting a NODATA response?
[I tried to look at the raw data to answer this question, but wasn't able to find out how to download it. I can see the probes that performed the measurement, but not that the results were. Am I doing something wrong?] Click on the bar that says 'Results' at the very bottom of this page? You should be able to download all the results and see the last 100
On Thu, Oct 11, 2012 at 9:07 PM, Andreas Strikos <astrikos@ripe.net> wrote:
[I tried to look at the raw data to answer this question, but wasn't able to find out how to download it. I can see the probes that performed the measurement, but not that the results were. Am I doing something wrong?]
Click on the bar that says 'Results' at the very bottom of this page? You should be able to download all the results and see the last 100
Wow, that's really hard to discover. Thanks.
On 11.10.2012, at 14:49 , Lorenzo Colitti wrote:
On Thu, Oct 11, 2012 at 9:07 PM, Andreas Strikos <astrikos@ripe.net> wrote:
[I tried to look at the raw data to answer this question, but wasn't able to find out how to download it. I can see the probes that performed the measurement, but not that the results were. Am I doing something wrong?] Click on the bar that says 'Results' at the very bottom of this page? You should be able to download all the results and see the last 100
Wow, that's really hard to discover. Thanks.
Isn't making that more obvious already very long on the to-do-list?
On Thu, Oct 11, 2012 at 8:56 PM, Lorenzo Colitti <lorenzo@google.com> wrote:
When you did the AAAA measurements, did you distinguish between the probe not getting a response (e.g., if the dig command timed out) and the probe getting a NODATA response?
Ok, after having done my own analysis, it looks like the answer to that was no. Strictly speaking, AAAA filtering is when the recursive resolver returns NOERROR with zero records even if there was an AAAA record. However, this but this rare in the raw data here. Many of the cases that the post reports as filtering seem to be problematic home routers returning "not implemented", timing out, or returning SERVFAIL when asked for AAAA records. While this is still a problem for IPv6 (especially in the case of timeouts) it's not really AAAA filtering. The "not implemented" seems to be mostly to blame for the failures in the French AS at the top of the list: Results (produced by the attached Python script): A queries (udmlog_1004102_2012_39.txt): Results: {'successful': 167183, 'norecurse': 382, 'ERR_timeout': 5722, 'refused': 1163, 'ERR_senderror': 416, 'server-failure': 362} AAAA queries (udmlog_1004103_2012_39.txt): Results: {'filtering': 420, 'successful': 165667, 'norecurse': 383, 'ERR_timeout': 6057, 'refused': 1156, 'not-implemented': 774, 'ERR_senderror': 415, 'server-failure': 317} Filtering DNS servers: 124.102.51.104: 96/98 (98.0%) 79.131.194.22: 95/95 (100.0%) 124.212.215.50: 97/97 (100.0%) 88.188.129.85: 2/95 (2.1%) 133.38.4.79: 96/96 (100.0%) 2.85.23.226: 32/55 (58.2%) 81.56.47.149: 2/96 (2.1%) Filtered probes: 2299: 96/98 (98.0%) 2848: 2/96 (2.1%) 3028: 95/95 (100.0%) 2896: 96/96 (100.0%) 2891: 97/97 (100.0%) 2197: 2/95 (2.1%) 268: 32/55 (58.2%)
participants (4)
-
Andreas Strikos -
Daniel Karrenberg -
Lorenzo Colitti -
Vesna Manojlovic