PCH peering survey 2016

Bill Woodcock woody at pch.net
Thu Sep 15 00:04:17 UTC 2016


Five years ago PCH conducted the first, and to date only, comprehensive survey characterizing Internet peering agreements.

The document that resulted can be found here: https://www.pch.net/resources/Papers/peering-survey/PCH-Peering-Survey-2011.pdf <https://www.pch.net/resources/Papers/peering-survey/PCH-Peering-Survey-2011.pdf>

That document was one of the principal inputs to an important document that the OECD publishes every five years, one that recommends communications regulatory policy to OECD member nations: http://www.oecd.org/officialdocuments/publicdisplaydocumentpdf/?cote=DSTI/ICCP/CISP(2011)2/FINAL&docLanguage=En <http://www.oecd.org/officialdocuments/publicdisplaydocumentpdf/?cote=DSTI/ICCP/CISP(2011)2/FINAL&docLanguage=En>

The survey had several useful findings which hadn’t previously been established as fact—most notably the portion of peering relationships that are “handshake” agreements, without written contract. These findings have improved the regulatory environments in which many of us operate our networks.

At the time of the 2011 survey, we committed to repeating the survey every five years, so as to provide an ongoing indication of the direction peering trends take. It’s now five years later, so we’re repeating the survey.

The survey is global in scope, and our goal is to reflect the diversity of peering agreements in the world; we’re interested in large ISPs and small ISPs, ISPs in Afghanistan and in Zimbabwe, bilateral agreements and multilateral, private and public. Our intent is to be as comprehensive as possible. In 2011, the responses we received represented 86% of all of the world’s ISPs and 96 countries. We would like to be at least as inclusive this time.


In 2011, we promised to collect the smallest set of data necessary to answer the questions, to perform the analysis immediately, and not to retain the data after the analysis was accomplished. In that way, we ensured that the privacy of respondents was fully protected. We did as we said, no data was leaked, and the whole community benefited from the trust that was extended to us. We ask for your trust again now as we make the same commitment to protect the privacy of all respondents, using the same process as last time. We are asking for no more data than is absolutely necessary. We will perform the analysis immediately upon receiving all of the data. We will delete the data once the analysis has been performed.

The Survey:

We would like to know the following five pieces of information relative to each Autonomous System you peer with:

• Your ASN
• Your peer’s ASN (peers only, not upstream transit providers or downstream customers)
• Whether a written and signed peering agreement exists (the alternative being a less formal arrangement, such as a "handshake agreement")
• Whether the terms are roughly symmetric (the alternative being that they describe an agreement with different terms for each of the two parties, such as one compensating the other, or one receiving more or fewer than full customer routes)
• Whether a jurisdiction of governing law is defined
• Whether IPv6 routes are being exchanged (this year, we’ll still assume that IPv4 are)

The easiest way for us to receive the information is as a tab-text or CSV file or an Excel spreadsheet, consisting of rows with the following columns:

Your ASN: Integer
Peer ASN: Integer
Written agreement: Boolean
Symmetric: Boolean
Governing Law: ISO 3166 two-digit country-code, or empty
IPv6 Routes: Boolean

For instance:

42 <tab> 715 <tab> false <tab> true <tab> us <tab> true <cr>
42 <tab> 3856 <tab> true <tab> true <tab> us <tab> true <cr>

We are asking for the ASNs only so we can avoid double-counting a single pair of peers when we hear from both of them, and so that when we hear about a relationship in responses from both peers we can see how closely the two responses match, an important check on the quality of the survey.  As soon as we've collated the data, we'll strip the ASNs to protect privacy, and only the final aggregate statistics will be published. We will never disclose any ASN or any information about any ASN. We already have more than 8,000 ASN-pair relationships documented, and we hope to receive as many more as possible. We'd like to finish collecting data by the end of September, about two weeks from now.

If you’re peering with an MLPA route-server, you’re welcome to include just the route-server’s ASN, if that’s easiest, rather than trying to include each of the peer ASNs on the other side of the route-server. Either way is fine.

If all of your sessions have the same characteristics, you can just tell us what those characteristics are once, your own ASN once, and give us a simple list of your peer ASNs.

If your number of peers is small enough to be pasted or typed into an email, rather than attached as a file, and that’s simpler, just go ahead and do that.

If you have written peering agreements that are covered by non-disclosure agreements, or if your organizational policy precludes disclosing your peers, but you’d still like to participate in the survey, please let us know, and we’ll work with whatever information you’re able to give us and try to ensure that your practices are statistically represented in our results.

If you're able to help us, please email me the data in whatever form you can. If you need a non-disclosure, we're happy to sign one.

Finally, if there are any other questions you’d like to see answered in the future, please let us know so that we can consider addressing them in the 2021 survey. The question about IPv6 routing in this year’s survey is there because quite a few of the 2011 respondents asked us to include it this time.

Please respond by replying to this email, before the end of September.

Thank you for considering participating. We very much appreciate it, and we look forward to returning the results to the community.

                                -Bill Woodcock
                                 Executive Director
                                 Packet Clearing House
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://mailman.nanog.org/pipermail/nanog/attachments/20160914/bf063e6a/attachment.pgp>

More information about the NANOG mailing list