The Days of the Peering Renaissance

Question

DrPeering - 


Is it me, or are things in the peering ecosystem changing rather quickly these days?


Astrid Levinson

Answer

Astrid -


I have seen this as well. Let’s look at a historical comparison. When the black plague swept Europe, it killed the pious equally as well as the populous. The renaissance that followed has been characterized by the populous questioning everything, not accepting purely on faith. This questioning of everything, this freedom from blind unquestioning acceptance of religious doctrine, led Da Vinci to meditate on questions that ultimately led to evolutions in insight and innovation.


The participants at the core of the Internet are questioning conventional dogmatic views of how interconnection should work in the U.S. ; they are considering other interconnect models. The Open-IX initiative inspires the European-model IXPs into Virginia and New York, providing a model of colocation-neutral peering-based traffic distribution. Remote Peering enables peering without physical colocation, and paid peering evolves from the emotional “extortion” characterization into a less emotional interconnection, maybe practical approach to getting content to the eyeballs. Interconnection is evolving.


And with all of these alternatives, the peering dogma rears its head. There is a “right” and a “wrong” way to interconnect, you will hear. These strongly held positions reinforce one’s existing emotionally-held positions; it is the path of least resistance. It is harder than  understanding the rationale for a blend of interconnect options that one can choose from. 


As a result, peering discussions get heated and personal. I always wondered why all of these peering coordinators fail to find an objective common ground. They latch onto their belief of “good” and “evil”, of motivations and intentions of the various players, and create a view and persuade others to that world view. Many enjoy the drama, and many appear incapable of understanding alternative perspectives and actively attack them. 


In reality, we may be witnessing that there are different Internet interconnection blends that, together, deliver the best performance characteristics for the specific network application.


The Path Forward


First we must accept the dichotomy between content and ISP. The content providers universally got into peering to improve the end user experience viewing their pages, where ISPs primarily focus on peering to minimize their cost of delivering bits. So accept the different motivations (end-user experience vs. cost of packet delivery), different units of commerce (pages vs. packets).


We also need to accept that there will be different levels of network expertise, and experience levels. ISPs have always employed peering, while many content and enterprises are just getting into peering. These new participants may need some education, some help, perhaps some introductions. It benefits everyone to grow the peering ecosystem.


Second, we need to understand the views and positions that various players have adopted. 

These peering evolutions are taking place in the context of a peering community with views spanning a broad spectrum of peering philosophies. 


A Spectrum of Peering Philosophies


Here is an image describing a few of the peering philosophies that I have seen over the past twenty years in the peering sector.

The Interconnectionalists


I first met the “interconnectionalists” in the late 1990’s when they resisted the word “peering.” These folks wanted and still want the word banished. They prefer the word “interconnection” to describe the directly attached network adjacencies. To them interconnection is akin to sex - a private act between consenting networks. The relationships are bilateral, not public, and they know who are desirable interconnect partners. The problem they have with “peering” is that interconnection is business, and peering brings with it emotional turmoil (“what really makes you a peer? scale, scope? user base? market share? we are good in different ways than you.... etc.) and that makes it personal and then, the victim makes it public.


The Peering Taliban


On the other end of the spectrum are what I call the “Peering Taliban.” The Peering Taliban are a more vocal and fervent bunch who hold tight onto the views of the 2000-era Internet. At this time, remember that the U.S. peering model was transitioning from carrier-owned IXPs to carrier-neutral IXPs. The “big bad telcos” were “bilking” ISPs for both expensive circuits and expensive ATM ports, and no one liked paying their telco-competitors for anything. The next best alternatives at the time were point-to-point metro circuits, which were even more expensive. 


Faced with only these bad choices for traffic exchange, the community of ISPs and large content providers rallied behind emerging carrier-neutral IXPs (Equinix, Switch and Data,. Terremark a little later, etc.). Over time, carrier-neutral colocation centers operating their own ethernet-based IXPs, emerged as a cost effective open market for free peering and transit purchases. The community “won” with ethernet (a technology that they loved), carrier-neutrality (they didn’t have to reinforce a competitor), and a business case for peering that everyone benefited from as the peering population grew. Together, these IXPs and the peering community embraced the aligned community interests era of interconnection, and the IXPs supported this “love in” in the form of peering forums, a place on land or sea for aligned interests to meet and get interconnected. This became the peering religion in the 2000-2010-era Internet.


But these forums were owned and operated by a self-selected set of IXPs with a vested interest in maintaining this physical presence model of interconnect. In the U.S., there was no interest in evolving a broader set of interconnection alternatives that may not require the IXP services.


For some, the pendulum swings from side to side between these extremes, mostly resting in the middle as “Peering Pragmatists.”


The Peering Pragmatists


The “Peering Pragmatists” are only focused on what is in their immediate self-interest. They will be the last to join the masses flocking to a new interconnection technique unless they see immediate benefits. They won’t build into an exchange point until there is enough traffic to peer for free to cover the cost of peering. They don’t care so much about whether a competitor or telco gets paid for peering services, only that these peering services save them money or helped them get deals. They don’t care about new carrier-neutral IXPs unless they had critical mass and immediate value. A stand-alone business case is required to warrant building in. There is no “leap of faith” for these organizations. These folks will embrace the broad blend of interconnection and distribution options (Transit, Partial Transit, Peering, Paid Peering, Remote Peering, Distance or “cloud peering”, CDN, Caching etc.) that are relevant and immediately helpful.

Summary

You are peering during the most exciting time since the commercialization of the Internet. When you speak with peers, you will hear views espoused across the spectrum of “interconnectionalists, ”Peering Pragmatists,” and “Peering Taliban.” There is some validity in each of these perspectives, and identifying the camp will provide some insights and help you avoid saying the wrong thing when negotiating peering during this rapidly evolving ecosystem. I believe that ultimately, over time, most in the Internet Peering Ecosystem will tend towards the center.