Content with an edge

As Akamai's content delivery network evolves, 
can it keep carriers from taking too much interest?

Sometimes it takes a half-naked woman to show the world what it is missing. In January 1999, the Web site of the popular lingerie retailer Victoria's Secret hosted a heavily advertised, well-attended fashion show Web cast that was notable not for what the models were (or weren't) wearing, but for the large number of complaints it generated from users who found the images and audio too slow to load.

Slow Web site performance was nothing new back then, and in fact, visitors to many popular sites had come to expect it, but those Web cast problems considerably raised the profile of poor Internet performance and reliability. The ongoing question “How do I get Web pages to load faster?” suddenly was worth answering for network operators, ISPs and a fledgling group of content delivery networks (CDN) and caching companies.

Cambridge, Mass.-based Akamai Technologies was one of the first CDN firms around and now owes most of its success to its ability to answer that one simple question more quickly and more directly than anyone else. Akamai's response was to install servers at the edges of public networks that would cache frequently accessed Web elements, refreshing them occasionally so the most desired content on the Web was that much closer to the users who wanted it.

To ensure performance, Akamai ran the servers itself, effectively becoming a franchise content network within the boundaries of carrier public networks. Its push into those networks has occurred with stunning quickness: When the company reported second quarter earnings late last month, it announced it now has 11,600 content servers deployed in more than 820 networks around the world.

“A key predicate of our approach is the amazing dispersion of users across the networks that make up the Internet,” says George Conrades, chairman and CEO of Akamai. “No single service provider has more than 6% of the Internet's access bits. If you want to create a performance enhancement for the Internet, it has to be done at the edges of those networks. As a content delivery service provider, you have to be in all those thousands of networks.”

Adds Jonathan Stefansky, vice president of network infrastructure for Akamai, “We want to get into all 7000 [carrier and ISP] networks a server in front of every eyeball.”

However, while content delivery companies such as Akamai, CacheFlow, Digital Island and InfoLibria have proved important midwives for delivering static content to users more effectively, the CDN is moving well past its original reason for being. In the future, CDNs may address many aspects of creating and managing content, taking this burden away from network operators that want to offer content services but do not want to get bogged down in the details. Akamai, in that scenario, could become something more of a wholesale content service provider than the technology vendor it traditionally has been.

Content may shift

Akamai's original service, FreeFlow, focused on speeding delivery of static content elements from frequently-accessed Web pages to end users. Its latest set of solutions, EdgeSuite, expands into an array of new content functions, including more efficient content creation, object delivery and streaming media development, content monitoring, content reporting and content targeting.

Also, Akamai is driving the development of a new content delivery standard, the Edge Side Includes (ESI) content assembly language. ESI is a dynamic markup language that helps Web site developers determine which cached site elements need to be rebuilt as they are accessed. For example, ESI could be used to dynamically develop targeted Web page ads as different types of users with differing needs and demands access these pages.

“We want to get ESI's ecosystem to the point where there is a strong developer community working on it,” says Keiran Taylor, director of product management for Akamai. Akamai also is contributing to the development of XSLT, an ESI interface to XML that will allow ESI-treated pages to be easily rendered in HTML, WAP or whatever other language needed to reach wireline and wireless users on different kinds of networks.

Development of new content services and introduction of further flexibility in content creation eventually points toward the personalization of Internet content based on customers' unique interests, bandwidth and browser requirements and local traffic conditions.

While Akamai steps up CDN technology, it also is responding to how the business of content is changing. The AOL Time Warner deal renewed the possibility that network operation and content ownership can happen under the same roof an idea long under consideration in the telecom arena but never fully realized. Also, service providers such as telcos and ISPs have begun to feel pressure to diversify their service portfolios and improve the performance and reliability of their Internet and content services.

Meanwhile, operators of Web sites, which had the most to gain from Akamai's ability to cache content and speed it to users' screens, are dying off in droves during the worst economic downturn in more than 10 years. For a specialty service provider like Akamai, the loss of audience could have been debilitating. Conrades has admitted that the downturn was responsible for a larger-than-expected loss for the fourth quarter of 2000 as well as the mass layoff of 200 employees announced in April. However, despite these sour notes, Conrades recently said Akamai has been only “marginally affected” by the downturn.

“The market landscape has shifted because the dotcoms didn't execute,” Conrades says. “We're in a good position because we have server robustness that can be directed at content assembly and other services, rather than just processing Web pages at the network edge.”

Chris Schoettle, chief operating officer for Akamai, says the job cuts were necessary because Akamai's organization had grown so rapidly in the year before the downturn occurred. “We went from 400 people to 1400 in a year, and there has to be some digestion on the other side of that,” Schoettle says. “We're fighting trim at 1200 people, and the layoffs will help us stay on a track toward positive EBITDA.”

Cache and carriers

That particular issue just when Akamai will be positive EBITDA has been a source of growing concern for financial analysts, who figured the company had been thrown off schedule by market woes. However, in announcing second quarter 2001 earnings results last month, Conrades reassured Akamai watchers: “We planned to be positive EBITDA in the second quarter of 2002 and we will still be positive EBITDA in the second quarter of 2002. Our current business plan is well-funded, and we have a low-cash expenditure model.”

Part of the reason Akamai has a low-cash model is that it maintains only a network of servers, not pipes. Carriers are the ones with the bandwidth, and Akamai is taking advantage of those assets, as well as carriers' inability or lack of desire to manage the content in their networks.

In the past, carriers were unsure whether to view CDNs such as Akamai as a competitive threat. On the one hand, Akamai was improving the carriers' overall bandwidth efficiency, but it was simultaneously speeding end users' access to content, something carriers could not seem to accomplish themselves. “Carriers have looked at Akamai and have been fearful,” says Joel Yaffe, senior industry analyst with Giga Information Group.

Typically, Akamai has co-located its servers in carrier networks, meaning its servers have occupied carrier resources; by selling services to end users who were also likely to be customers of the carrier, Akamai effectively undermined the potential for the carrier to provide those services itself. More recently, public network operators have started partnering with Akamai to become resellers of its services, meaning Akamai not only would co-locate servers in a carrier network but also would manage an entire portfolio of content services for the carrier.

Though resale does offer carriers a chance to brand their own content services, both of Akamai's partner models co-location and resale put carriers in a position in which they outsource, to a large degree, control of the content riding on their networks. So should carriers be fearful? The ones that are signing up to be partners, such as XO Communications and Canada's Group Telecom, do not seem too worried about relinquishing content control to the Akamai juggernaut.

“It's a good thing,” says Steven Koles, senior vice president of marketing for Group Telecom. The carrier even calls its service GT Akamai. “We're associated with a leading brand name for content and we think our customers can appreciate that we stick to our knitting.”

“Carriers have reconsidered the investment that they need to make in the content business,” says Conrades. “Network operators are not technology development companies and they can outsource that function.”

Koles says Group Telecom always considers whether it is smarter to build or buy a new capability into its network, “but with this, we didn't think about it too much. We just looked at all the different content delivery solutions and went with Akamai.”

Schoettle, a onetime executive at AT&T WorldNet and Lucent Technologies, says carriers do not have as big of a “Not Invented Here” complex as they once did. “They have come to realize that sometimes partnering for a best-of-breed solution is the way to go,” he says. “They look at us not as an overlay that needs to be deployed on the network, but as part of the Internet itself.”

Ultimately, Koles says reselling Akamai gives Group Telecom service the breadth it really needs to compete effectively. “Content delivery and other content services are very sticky offerings, unlike something like long distance,” he says. “We've seen a lot of traction among media companies that want these services for their Web sites.”

Stefansky says Akamai wins, too, in its resale relationships with carriers. The stable telcos give Akamai a sense of security it can no longer find selling directly to the dotcoms. “Telcos make attractive partners because they have been around for a long time and the assumption is that they will [stay] around for a long time, and they have a lot of eyeballs,” he says. “The beauty of having a distributed network of servers is that we don't depend on whether the dotcoms are doing well or not. Being in networks provides us with a strong ROI position.

'Telcos make attractive partners because they have been around for a long time and the assumption is that they will [stay] around for a long time.' --Jonathan Stefansky vice president of network infrastructure

Suite deal  

About 17% of Akamai's revenue comes from its reseller channel, according to Taylor. And in terms of total market share, Akamai owns an overwhelming 75% of the CDN market. Its new EdgeSuite is already making an impact with 30 customers signed so far, about half of them upgrading from FreeFlow.

One of those customers is the aforementioned Victoria's Secret. Though Akamai was not around to save the company from its Super Bowl Sunday disaster, Victoria's Secret has been a customer since June 1999 and invested in EdgeSuite two months ago to support the creation of new content and commerce transactions for the site. That progression from a single content application to multiple content services is exactly what Akamai is looking for as the content business shifts and the CDN landscape changes with it.

'We're fighting trim at 1200 people, and the layoffs will help us stay on a track toward positive EBITDA.' --Chris Schoettle chief operating officer

If the AOL Time Warner deal got network operators to start thinking about content again, then Cable & Wireless may have been one of the first carriers to express its intention to own a piece of the content market the telco acquired Digital Island last spring, an aggressive move that goes far beyond the resale opportunity that Akamai is offering carriers. With Digital Island, C&W is investing not only in the edge of its network, but in a competitive edge owning the tools and services that will introduce better performance and reliability for its Web site operator customers, Internet users and streaming media users.

The execution of such a deal would seem to suggest market-leader Akamai might be next. Conrades does not think so. “For Akamai, a large network operator acquiring us would really dampen our effect. Besides, they would have to pay an awful lot of money for us, and we're not for sale.”

However, Giga's Yaffe says Akamai will have its hands full trying to remain independent while continuing to invest in its evolution as a content service provider. Though analysts still credit Akamai management with running their business effectively, the gasping economy has driven the company's once-high-flying stock price into the gutter (see figure on page 22-23).

Conrades asserts Akamai has plenty of funding to carry it through to positive EBITDA, but that probably will not stop bids coming in from carriers that see the chance to own a lucrative content franchise that could make them money from the traffic on other carriers' networks.

That possibility seems so compelling, it is surprising that carriers have not invaded the CDN business sooner. Their own self-centered view could have been part of the problem. While carriers have been busy building broadband pipes to carry more traffic, they have simply ignored the nature of the Internet, where content has traditionally resided, and how their users might benefit if they sought to change that model themselves.

“Carriers were not thinking about their customers,” says Yaffe, and the implication that Akamai has been thinking about them hangs in the air. “The real question is not how Akamai does this,” Yaffe says, “but why carriers have never figured out how to do it themselves.”