Keep it local
The business model for content delivery networks is changing fast, finds Tim Phillips, with effective local coverage looking like it could be one of the possible answers to a successful future in the market.
When Capacity wrote about content delivery networks (CDNs) in 2009, the industry was predicting consolidation: we observed the amount of venture capital dedicated to building CDNs in the boom was greater than the size of the market, and the price those would-be CDNs could charge for delivering video was dropping fast. The predictions of our experts came true. Smaller CDNs have been snapped up or are finding a new business model, and prices for the basic job of delivering content continue to decline. While video continues to mushroom – HD is four or five times the number of bits to deliver as conventional television – there’s little additional revenue from doing the extra work.
That has led many CDNs to accelerate their diversification and create new types of business. While those services are a lifeline for established CDNs, there’s still the problem that soon up to 90% of legitimate internet traffic will be video, and that users of that video – increasingly paying through subscription services – are becoming more demanding. If CDNs are unable to guarantee quality, or the various providers in the delivery chain are not being paid to assure that quality, then neither content providers nor customers are getting what they pay for.
Dan Rayburn, an analyst at Frost & Sullivan who closely monitors CDNs through his blog businessofvideo.com, points out that the pure-play CDNs have grown to rely on the value-added component. “Akamai, for example, gets 55% of its revenue from value-added services. Limelight gets 40% from the same route,” he says. Rayburn calculates that the pricing for video delivery declined by between 40% and 45% in 2009, and will fall by between 20% and 25% in 2010.
A difficult environment
In a difficult financial environment, when existing CDNs have been starved of investment while having to keep developing services, those value-added services have helped to dampen the volatility of the business. “The average contract length in the CDN business is 12 months. There’s a lot of jumping around and churn,” says Rayburn. “There is also a false perception that the CDNs grab big customers with low cost, but they can’t afford to do that.”
But there is an alternative to the proliferation of global “value-added” CDNs – the possibility of developing a more geographically-restricted CDN offers content providers a guarantee that the service they want to deliver (most obviously, video) will be delivered with an assurance of quality.
One of the smaller CDNs which has evolved into this business by degrees is Velocix. Originally created as a start-up in the CDN boom of 2002, the company was acquired by Alcatel-Lucent in 2009. Since then it has concentrated on creating the products which can help service providers deliver video to their customers. Its best-known public customer in this is Verizon, signed in 2008. At the time, Velocix chief marketing officer John Dillon said that “most CDNs today have worldwide networks but thin capability region-to-region.” It has since signed a similar European deal with mobile operator TalkTalk.
Changing business model
In the broadcast world, global coverage is not yet an advantage: video content is protected by copyright, so it is more important to deliver it at a high quality to subscribers in your home region than to deliver it to non-subscribers worldwide. At the other end of the value chain, a service provider is picking up the cost of delivering data to subscribers, but the CDN is being paid for it.
Anthony Berkeley, marketing and business development director at Velocix, claims this business model has become more attractive since the Verizon deal was signed. “In the last 18 months the CDN world has changed. So there’s how business was done in the past, and how business will be done in the future. We realised that we were moving away from business-like accelerating websites, towards an order of magnitude increase in demand,” he says.
The acquisition by Alcatel-Lucent signalled a change in strategy towards providing the building blocks of a local CDN, optimised for video, according to Berkeley.
“We have produced a product, hardened the software, purchased the infrastructure and implemented it. We see the emergence of a telco infrastructure.” That infrastructure, he says, will solve two problems. The first is that content providers will be able to pay service providers for delivery; the second is that the service providers can create their own peered network of CDNs.
“Service assurance isn’t going to happen outside the service provider network. Content providers thought they were paying for delivery – but they realised they were paying the middleman. As the model shifts from moving files to moving video, it becomes obvious that what they pay for doesn’t ensure the quality they want it to,” Berkeley explains, “Imagine a network of CDNs with the same structure, a global network with industry-standard components. We get better quality, pushed right to the edge.”
BT’s 2011 CDN launch is targeted at a similar customer. It will be sold as an add-on for service providers who want to be part of the 21CN launch: they can simply tick a box to also use the capability of BT’s wholesale CDN in the UK. With 20 PoPs in the UK, “which is four or five times the number of PoPs of any other CDN,” in the words of Simon Orme, strategy director for content services at BT Wholesale, “we can report accurately on demand and geographical traffic patterns, which helps the service providers plan their service. And it is integrated with the underlying network, which means assured quality.”
When BT evaluated the CDN market in 2008, it reached a similar conclusion to the management of Velocix: that the pure-play CDN model was less relevant for broadcasters and ISPs than a local infrastructure.
“We saw two things,” says Orme. “First, the role of the pure-play CDN, if you wind it back for three or four years was as a useful arbitrage of international peering. Creating a peering relationship with an international CDN was very valuable for an ISP. It’s still valuable, but we’re seeing an explosion in local content, and so the pure-play CDN becomes less valuable in this context.”
BT was already noting the emergence of ISP CDNs from the larger service providers, and decided this was the next generation of CDN model. It helped that BT was building a next-generation IP network which could contain this model, and create a business on the back of it. While Velocix, owned by an equipment manufacturer, has “productised” its CDN to sell to ISP customers, BT wants to be a wholesaler. “We are establishing a CDN as a multi-tenant platform, and we will get a unique capability out of that. It means we will have a standardised content-delivery platform for the UK, with a homogenised, standardised way to reach the audience. We still give the tools to the ISP to differentiate themselves, but we are looking to find collaborative ways to create a service,” Orme says.
“They will be carrying the same volumes of traffic, but shipped at a lower cost than regular traffic if they are part of our CDN.”
While BT has no plans to launch a similar product overseas, it predicts that other large-scale networks will increasingly develop national capacity. AT&T and Deutsche Telekom, for example, have been aggressive in creating CDNs, but there have been few other entrants to the business.
Rayburn doubts that new CDN entrants will radically reshape the market in the short term. “There are smaller players in the industry still, but will they displace Akamai or Limelight or Level 3? No. The only company that might make the jump is AT&T. It takes years to build a business, and it’s not about what assets you have, it is whether you can execute.”
The problems of patchy execution might yet help local service providers to create a sustainable business. When Capacity spoke to Rayburn, he had just experienced a bruising week after having broken the story that Netflix, the largest provider of on-demand video streaming over the internet, had moved some of its CDN business to Level 3 from Akamai.
Rayburn was anxious to point out that the shifts in business from one CDN to another aren’t whims from the content providers, but the realisation that – in a market with low loyalty and huge risks for the content provider – a reflection that CDNs who don’t deliver create business problems for their content provider customers.
“What nobody seemed to want to write about is that Netflix was issuing credits to its customers for poor service. Investors get blinded by the information they get from these companies about growth, but no one wants to talk about what the customers are getting,” Rayburn says. “And if I’m a content provider, that is what’s important to me.”
The volatility in the CDN market, which has gone through two booms already inside a decade – coupled with the poor margins for pure-play CDNs in delivering video – might mean that the online video boom could be delivered increasingly by local service provider CDNs.
CDNs: from standalone to ISP
The traditional CDN model houses content off-net, but in geographically-dispersed servers. The benefits are improved performance for consumers and improved resilience in case of network problems. Long-established providers such as Akamai and Limelight partnered with content providers and have diversified to create a business that helps all types of websites run faster, more reliably and with better targeting for advertisements.
Global carriers have long been expected to create competition, but Level 3 is the only one which has created a dominant CDN business. Others, notably AT&T, have created their own on-net CDN offerings – but the large investment, small size of the market and low margins have slowed investment.
For the next generation of CDNs, especially those delivering video, it may be that ISPs and cable providers have a larger role. While they do not have global scale, they are able to offer more control over the quality of service to the consumer.
The value in value-added services
Whatever the threats to the profitability of their core business, diversification has protected the profitability and customer relationships of Akamai, Limelight and Level 3, which combine to serve at least two-thirds of the market.
Limelight has businesses in managed hosting, web development, interactive advertising, content management, DRM and e-commerce, as well as advertising: for example, in December 2009 it acquired Eyewonder for $110 million, which provides online campaign management. The businesses are complementary, not least because Eyewonder was previously paying CDN fees to Limelight.
Akamai was the first, and the most aggressive, diversifier. Today, for example, analyst Trefis estimates that online shopping content delivery accounts for around one third of its stock price – more than media content. Trefis adds that three out of four Akamai customers buy at least one value-added service, such as dynamic site acceleration or application performance.
In April 2010, EdgeCast Networks closed a $10 million financing from venture capital firm Menlo Ventures. It plans to use the funds to develop and launch value added services and expand its global network to meet surging customer demand. “Devices and content are coming online faster than ever,” said Alex Kazerani, CEO of EdgeCast, “and this investment will help us scale up to meet the demand that has surpassed even our most optimistic expectations.” EdgeCast won Capacity’s Best Market Innovation award in November 2009 in recognition of its successful global CDN and its agreements with Deutsche Telekom, Global Crossing and Dogan Telekom to offer partners an outsourced CDN solution.