A Buyer's Guide to Content Delivery Networks (CDNs)
Where have all the CDNs gone? None of the most prominent CDNs want to be thought of as CDNs anymore. What does that mean for buyers or for the industry?
Think of any two CDNs you know about. Now type "CDN" into Google. Notice something?
Or more to the point, is there anything you don't notice?
Almost as soon as CDN services become widely understood and the market adoption morphed from niche service to commodity, the companies with the largest market shares (I'm guessing at least one of the two you thought of above) have stopped identifying themselves with the sector they dominate.
When I penned this at the end of November 2012, CDNetworks, Akamai Technologies, Limelight Networks, and Level 3 Communications, which I perceive to be the most established CDNs, do not even feature on the first page of Google's results (aside from a reference to Akamai in a sponsored result for Rackspace). Level 3 leads the pack with an appearance halfway down on the second page.
The others do not appear within the first 15 pages of the Google search results.
Neither Akamai nor Limelight, nor most of the companies typically considered market-leading CDNs, seem to want to be thought of as CDNs. Indeed, currently it would seem to be cloud vendors that associate with Google's results for "CDN" early on and in the ad words, etc.
Six Tips for CDN Shoppers
- Do a traceroute/tracert from a client to the server that the stream is coming from and understand the number of hops in that trace. This is a fairly big subject, but understanding the traceroute/tracert command is essential to being able to see for yourself what your CDN is doing for you in terms of getting your content to viewers in the most efficient way.
- Ask about end-to-end latency for live streaming (the time it takes for the video input to the CDN to be output at the end user's player). Low latency (0.5s to 3s for RTMP or 8s to 10s for HTTP) is desirable, but it needs to be done right so that proxies and caches are well-placed in the network to handle load. When coupled with an understanding of traceroute/ tracert, you can quickly understand the topology of the CDN.
- Understand what live formats they support, since this will indicate a lot about their technology choices and platforms.
- Understand what players they support and how the players are configured. Be sure that the CDN's approach is something that you or your web team are comfortable with.
- Understand the API for the CDN's own workflow. If you want to integrate adverts, or prevent piracy, or manage subscriptions, this is absolutely crucial.
- Check the websites of the CDN. If the video on its front page doesn't work on whatever platform you choose to use to check its site with, or if it doesn't work the first time without any additional downloads, then do not do business with the company. Remember that this would have been your customer's experience of the video on your website had you decided to go ahead with it. It's a brutal, competitive market out there. Be tough.
CDNs Moving Up the Stack
My 2010 article "The State of the Stack" painted a picture of CDNs increasingly moving into their customer and supplier territories as a way to defend their core business. I would say that this has become so endemic among the "old" CDNs that they are at a risk of losing their identity as CDNs in the marketplace -- and failing to protect that core business after all.
In effect, by getting involved in production, protection, advertising, monetization, and similar, broader feature sets, these traditional CDNs risk losing their identity in the sector as the "go-to CDN service provider"-the pillar of business on which they are built. Also, by offering these other services as loss leaders in an attempt to aggregate CDN market share, they come into competition with their own customers, resulting in a war of attrition. With infrastructure as a service (IaaS) offers creating a very low barrier to entry for new players, this ends up simply making it unprofitable to run a business in this sector-not only for the old CDN (which typically has to maintain a lot of infrastructure), but also for the new entrants that have to risk a lot of capital to win market share.
This market distortion will ultimately leave the older CDNs one place to play: in the IaaS cloud themselves. Many have been offering platform as a service or software as a service since they were built (and could arguably be called the "original cloud vendors"), but until recently none have ventured into the IaaS space.
As Amazon has proven, there is more demand for CPU on demand than bandwidth on demand. With telecom operators now carrying their own CDN architecture, the networks increasingly manage traffic natively.
Overlay, Operator, Enterprise, and Boutique CDNs
Is your CDN provider an overlay CDN, an operator CDN, or an enterprise CDN provider? You'd be surprised how many people try to buy overlay CDN or operator CDN services to deliver into their corporate networks, when what they need is better architecture of their own within their enterprise. (That said, most overlays and operators will also have technology they can deliver into enterprises often as a managed service.)
Equally, if you want to deliver a webcast to a large public audience, then overlays and operators are both valuable, and your choice will vary with your target audience. If that audience is all "on-net" in the operator's telecom network, then you will get a far deeper reach and therefore greater efficiency and even service-level agreement guarantees that an overlay CDN cannot provide. Operators that were formerly telecom incumbents can offer deep services for applications such as IPTV, but only on-net.
Overlays can't do that, but they provide caches and relays in many locations close to high-speed interconnects/peering points with major ISPs, and often "inside" the largest ISP networks. So if you are delivering to a broad public audience who may, for example, be OTT or web TV viewers, then an overlay CDN is the right choice more often than not.
There are also "boutique" CDNs that operate like very well-operated CDN origin points; they take a single video stream and make it available directly to end clients using only high-speed internet connectivity at locations such as those used by overlay CDNs for their interconnection. These CDNs, while being centralized and thus an anathema to the normal concept of a distributed content delivery network's "edge caching" architecture, generally don't have any problem delivering to their local market, addressable through the internet peering at the peering point(s) they locate at. Also, by virtue of having a smaller, more closely run architecture, they can provide to their clients an attention to detail that larger providers with global aspirations cannot deliver. Some of these may even be entirely virtual-a well put-together "wrapper" on Amazon's or Rackspace's IaaS infrastructures.
The Perils of Repositioning
If Akamai and Limelight don't want to be thought of as CDNs, that begs three questions: 1) Is this a bad sign for the CDN industry? 2) How should we think of these former CDNs? and 3) Do they have a future?
In answer to my first question, I would compare this to BP's rebranding once it stopped being a U.K. company and changed its name from British Petroleum and adopted the slogan "Beyond Petroleum." I can't really imagine BP without thinking about petroleum. I can't think about Akamai without thinking about a CDN.
But my 6-year-old daughter may not grow up to instinctively think about BP when she wants to fill her car up. She may not even know BP produces petrol if it continues to bring her up thinking it is "beyond petrol" -- and that will be bad for its petrol business.
And so while Akamai, Limelight, and others try hard to distance themselves from "CDN" as a descriptor, attempting instead to be all things that their potential customers need them to be, new potential customers looking around the market for CDN services may not pick up the phone. This increasingly erodes market share.
In some ways, this is a good thing for the market -- creating opportunities for smaller, more focused, and more aggressive CDNs -- and this will lead to increased innovation.
However, while the larger players attempt to reposition, that search for the most valuable schism to dominate is tricky. While these providers are in an excellent position to offer IaaS services, they would also be perceived to be aping Amazon and Rackspace and other organizations that are leading in this space. Being late-comers to the market, their best hope would be to use existing relationships to upsell. But online publishers -- the core market for CDNs -- are a different market than the enterprise computing market that IaaS vendors typically target. I'm not sure potential customers will warm up to thinking of Akamai as a "cloud provider."
So only these organizations will ultimately be able to decide how they reposition themselves, if indeed my conjecture is right and that is what they are doing.
In answer to the final question, "Do they have a future?" certainly this repositioning coincides with a significant change in these companies' share values. Many of the companies I mentioned are trading at far lower prices than they were at the start of 2011, and in those 18 months, the industry has essentially entered a period of identity crisis.
I think it's important, though, to recognize that "content delivery networks" or "content distribution networks" do perform a fairly identifiable service: They enable large audiences to receive content efficiently and effectively. So it remains clear to me that to walk away from a close identification with these core service offerings and values is to move into the unknown.
For those of you who have picked up Streaming Media magazine hoping to choose a CDN to service your content delivery/distribution needs, you have a tougher job than ever because those that know how to offer CDN services are not necessarily making themselves seen. I don't envy your job of filtering through all the supplier options if you are new to the space. I've provided some tips in the sidebars, and most of the points from my buyer's guide in the 2012 Sourcebook still apply.
This article appears in the forthcoming 2013 Streaming Media Industry Sourcebook.
No one said choosing a CDN is easy, so here's a guide to the decision-making process. Don't miss our giant CDN flow chart.