Dear fellow networkers, I need your help! For the good of PeeringDB I am researching the accuracy of the current PeeringDB data set. We plan to compare three sources of information: peeringdb itself, publicly available listings from IXP operators ... and the ultimate source of truth: user submitted information, e.g. your "show bgp sum". Why? I'd rather trust 10 sightings in the wild than one entry in PeeringDB! :-) What can you do? ---------------- We've created a webapp where you can copy + paste the output from your routers' show ip bgp sum / show bgp sum / show ipv6 bgp sum. The webapp extracts the ASN and remote IP for the sessions and store those after your confirmation. Go to the following URL and submit your BGP data now! https://research.peeringdb.com/ If you prefer, you can also submit the data in CSV format [2]. What data are we using, exactly? -------------------------------- Only the following tuples of information are used: (remote_ASN, remote_IP) All other data is purged from the data set: I don't care if you are even exchanging prefixes or how many, nor does it matter what your own ASN is. The _only_ thing that matters is that you confirm that you have a BGP session up and running with a certain remote IP and ASN. You can submit such confirmations by copying + pasting your routers' bgp summaries. Please submit your BGP summaries from all your IXP facing routers! So when will I hear back about this? ------------------------------------ I will present the findings at the upcoming NANOG meeting in New Orleans [1]. Given that the NANOG meeting is approaching rapidly, I urge you to submit your data sooner rather than later. :-) Kind regards, Job Snijders [1] - CSV format should be formatted like column 1: ASN, column 2: remote IP, separated by a comma. example: "5580,195.69.144.229" [2] - http://www.nanog.org/meetings/abstract?id=2140
My dear fellow networkers, Good news everyone, 99% of the parsable data in PeeringDB is valid! :-) Measuring this number would have been inpossible without all the submissions to the research app. Thank you! If you are interested in the details, please see these slides: http://nanog.org/sites/default/files/wed.general.peeringdb.accuracy.snijders... Kind regards, Job On May 23, 2013, at 12:28 PM, Job Snijders <Job.Snijders@atrato.com> wrote:
Dear fellow networkers,
I need your help!
For the good of PeeringDB I am researching the accuracy of the current PeeringDB data set. We plan to compare three sources of information: peeringdb itself, publicly available listings from IXP operators ... and the ultimate source of truth: user submitted information, e.g. your "show bgp sum".
Why? I'd rather trust 10 sightings in the wild than one entry in PeeringDB! :-)
What can you do? ----------------
We've created a webapp where you can copy + paste the output from your routers' show ip bgp sum / show bgp sum / show ipv6 bgp sum. The webapp extracts the ASN and remote IP for the sessions and store those after your confirmation.
Go to the following URL and submit your BGP data now!
https://research.peeringdb.com/
If you prefer, you can also submit the data in CSV format [2].
What data are we using, exactly? --------------------------------
Only the following tuples of information are used:
(remote_ASN, remote_IP)
All other data is purged from the data set: I don't care if you are even exchanging prefixes or how many, nor does it matter what your own ASN is. The _only_ thing that matters is that you confirm that you have a BGP session up and running with a certain remote IP and ASN. You can submit such confirmations by copying + pasting your routers' bgp summaries.
Please submit your BGP summaries from all your IXP facing routers!
So when will I hear back about this? ------------------------------------
I will present the findings at the upcoming NANOG meeting in New Orleans [1]. Given that the NANOG meeting is approaching rapidly, I urge you to submit your data sooner rather than later. :-)
Kind regards,
Job Snijders
[1] - CSV format should be formatted like column 1: ASN, column 2: remote IP, separated by a comma. example: "5580,195.69.144.229" [2] - http://www.nanog.org/meetings/abstract?id=2140
-- AS5580 - Atrato IP Networks
On 13/06/2013 17:48, Job Snijders wrote:
Good news everyone, 99% of the parsable data in PeeringDB is valid! :-)
you mean: 99% of the parsable data in PeeringDB which is maintained by people conscientious enough to provide the output of "show bgp sum" from their routers, is valid. Good talk, and interesting research, but I have a feeling that there might be some unintentional researcher bias creeping in there. :-) Nick
participants (2)
-
Job Snijders -
Nick Hilliard