For some years now I have thought that the CDN world at some point would find a way to interconnect with itself. By that I mean that until now, each content delivery network lives in isolation on top of an IP ocean, exchanging nothing directly with its neighbors. Yet the idea of seamless caching and delivery across multiple cooperating CDNs is very attractive from a design standpoint, and I know that many on the bandwidth side have seen it in the future for a long while. Just in the last few days though, I have heard the emerging name for this, the 'Federated CDN', start to make real noise.
Dan Rayburn posted a video from the Content Delivery Summit where Cisco's Chris Osika discussed several potential business models (the relevant section starts at the 14th minute). And secondly, Jet Stream's Stef van der Ziel offered up a blog post describing an actual pilot implementation of between European CDN StreamZilla and an unnamed 'American telco' (wish I knew who!). And the term came up during a recent conversation I had with someone from NTT America as well.
The parallel with transit and peering - In a way, this is what we don't want to see: a tiered system with economic underpinings whose disagreements can turn into network partitions and whose relationships are too often determined by power and exclusivity. Nevertheless, it's a clear parallel from the network side. The internet is made up of discrete networks who exchange traffic with each other, sometimes via paid transit, sometimes via settlement free peering, and these days often with other paid-peering varients between the two. Together they go everywhere, even though individually they never did and never will. For the equivalent to work for CDNs, you have to do more than exchange traffic though. Having big enough pipes between two CDNs is less difficult than accounting for the cost/benefit achieved by each side in a transparent manner. But it ought to be doable, it'll just take a whole lot of elbow grease.
The beautiful flaw in the analogy - Unlike in the transit/peering world, CDN networks do not have a monopoly on reaching the endpoints their networks connect to. Every CDN can reach every point on the internet, it's just a matter of how they choose to do it. If two CDNs don't agree to work together, they can work around each other and still deliver the bits, if via no other means than by defaulting to a transit connection. As Akamai has shown, you can build out a heck of a CDN without building out any fiber, let alone last mile fiber. Thus, the only way for the last mile providers to leverage their position is to participate in a CDN federation, which could then act to dissipate the difficulties they face in handling the growth in video traffic.
The ecosystem - What I like about the Federated CDN concept is that it might provide a platform for so many different business models beyond today's limitations. The RLEC who has no interest beyond its territory and isn't big enough to draw interest from content providers could overlay CDN to sell access to even while cutting its own costs. CDNs with general regional expertise, particular fiber footprints, no fiber footprint but strong managed services capabilities, large distributed CDNs who still can't go everywhere, ILECs who want to provide a 'fast lane' without using that name or simply to manage costs from last mile video - all seem to work better in a federated CDN marketplace than they do (or don't) today.
Honestly, the only one that stands to lose is Akamai, only because they have so much marketshare that any change to a market structure they dominate is by definition a risk to that dominance. But even with Akamai, what you see these days from them is a move upmarket into value added services and a de-emphasizing of the more commoditized parts of their business. That commoditization of the basic CDN building blocks is something their actions show that they are already acknowledging and adjusting to if not preparing for. They're not fighting a holding action.
So it seems like everyone would benefit by integrating the CDN much more tightly with the internet via federations and/or exchanges. But it's still just a concept, with large scale adoption awaiting standards that don't yet exist. Still five years away?
If you haven't already, please take our Reader Survey! Just 3 questions to help us better understand who is reading Telecom Ramblings so we can serve you better!Categories: Content Distribution