United States: Startups Seek to Disrupt Content-Delivery Business

Content delivery networks (CDNs) and their increasingly popular private-network brethren - "enterprise CDNs" (ECDNs) - are big business. But the established vendors are seeing the playing field expand, as hungry new upstarts move to, well, eat their lunch.

A content delivery network, according to Meta Group, is defined as "a centrally managed collection of storage devices (caches) strategically placed on a network to enable the staging of Internet content closer to end users."

The market is getting hotter, driven by such expanding enterprise needs as elearning and other rich-media applications. Meta Group says the number of companies using CDNs increased 100 percent in 2001, to 2500, and that by 2005 the number could reach 14,000. Gartner, meanwhile, says 80 percent of enterprises will use ECDNs by 2006, for such applications as video on demand, while Frost and Sullivan predicts a $4 billion market as soon as 2003. All in the name of better web performance, and faster access to content - and for the enterprise, that means especially at remote or branch offices.

Early CDN leaders, including Akamai, Digital Island, and a host of other outsourcers, plus equipment providers like F5 Networks, now have some new competition nipping at their ankles. Also facing the newcomers are early ECDN players like Cacheflow, Cisco, and Inktomi.

At least two recent startups have begun claiming radical cost savings and a much quicker ROI for enterprises wanting to implement private CDN solutions - by using their innovative, new *software-based* approaches.

While initial costs for even a modest traditional ECDN can be between $150,000 and $300,000, according to Meta Group, solutions from these new startups can be deployed for as little as $25,000, says one company.

"Rather than build out this whole huge new network," said Justin Chapweske, CTO of Minneapolis-based Onion Networks, "we look at what's already in the enterprise, all of those existing network resources, and leverage them to provide a content-delivery solution."

The "Ad-hoc-amai" Solution

Onion Networks says its end-to end solution allows companies to create ad-hoc CDNs, not only at much lower cost, but without the potential last-mile problems of traditional CDNs, in which connections between the cloud and the content consumer can go down or get congested. Its "WebRAID" software on the client side (which installs simply as a browser plugin) transfers content reliably between the storage and operating system - "very quickly," CTO Chapweske said.

"Its essentially RAID over HTTP," said Chapweske, "allowing us to transfer content from multiple web sites, multiple caches and peers on the network in parallel. So we get very fast, reliable downloads." He added that this WebRAID technology can be integrated into enterprise applications to provide fault-tolerant content delivery.

One of the unique things about the technology, he said, is that it allows for "parallel streaming" -- for example, to download video files from multiple points in parallel. If downloading from T1 lines in parallel, he said, it does super-fast downloads in the range of multimegabytes per second. And, Chapeweske added, you can view the content as it's being downloaded. "That's huge for training."

"WebRAID even allows cost-effective distribution of full-blown movies," Chapweske continued. "That's not something you can do with an Akamai system. It's just cost prohibitive."

In addition to the obvious cost advantages of this software-based approach, reliability is better than with a traditional CDN. "Because Akamai is a purely DNS-based solution, you have no way to deal with network errors or find other routes," Chapweske noted. WebRAID, on the other hand, discovers other replicas in the network, he said,
making content distribution more reliable. "It provides an ROI that's much quicker than a hardware-intensive, cache-based CDN."

Customers of Onion Networks to date (at least those they're allowed to name) include iGames and Jibe Inc. For the latter firm, it's creating a media delivery application specifically for video on demand.

A New Peer-to-Peer Generation

A second promising content-delivery startup, whose solution, like Onion Networks', also grew out of peer-to-peer (P2P) technology, is Kontiki of Mountain View, Calif., which was founded by former Netscape marketing guru Mike Homer. P2P got the world's attention after the enormous adoption of the Napster music-sharing system, and the technology has since spawned a number of companies, many now addressing the enterprise market.

Kontiki calls its technology "a software-based secure media delivery solution for enterprises." The company says its suite of products, which it calls a "delivery management system," improves the effectiveness of company communications and training to employees, partners, and customers. It provides enterprises with a suite of applications to help deliver, control, and track digital media securely and efficiently.

Kontiki says its technology makes it simple to deliver digital media directly to end users, enabling cost-savings and efficiencies across such enterprise functions as corporate communications, e-learning and training, marketing, and sales.

Even though an infinite supply of digital content is already available on the Internet today, according to Kontiki, there remains a bottleneck between unused capacity on the backbone and the well-equipped PCs now standard on most desktops, which is preventing digital media from becoming pervasive on the Internet.

Kontiki says its software-based approach removes many of these barriers. Its technology utilizes "the untapped bandwidth and storage of millions of networked PCs at the outer edge of the Internet," controlled by a series of redundant and fault-tolerant network management servers directly attached to the Internet backbone. Kontiki provides users with what it calls TV-quality content right to their PCs via "a simple, intuitive and hassle-free interface that allows them to obtain and manage the content they want, easily and automatically." That interface, however, and Kontiki's entire solution, is a standalone, proprietary software package. The client software is not a simple browser plugin.

The company released its first software in December 2001, later revising it in April of this year. Its downside, some say, is that it features a proprietary, branded GUI, which media companies (especially) don't like. And that may be one reason Kontiki now seems to favor the corporate enterprise market.

How fast is Kontiki adding customers and users? The company's being somewhat coy. It does list ten or more customers on its web site (some directly related to its investors), including Loudcloud, Verisign, CNet, Amazon, and MP3.com. In April, it said it had some 120,000 users in its network, largely individuals who download the client software from the company's site. A PR spokesperson said that number is now 185,000.

One Analyst's View

Can P2P catch on in the enterprise, beyond the well known success of at least one company, collaboration firm Groove Networks? "Because of network terrorist applications like Napster" says Meta Group senior analyst Peter Firstbrook, "it will take a while before P2P has a good image in most IT shops." Accordingly, he said, "we advise P2P vendors to get away from that term and call themselves 'distributed' or other more acceptable adjective." Firstbrook went on to say: "Caches have undergone more transformation with added tools for remote management, security, reporting, streaming content, and so forth in the last year. The new startups will help push this evolution, making the cache a more active network component than it has been in the past. New features like SSL acceleration, virus and content inspection, digital rights management, content creation, distribution, remote storage of non-HTML content, content synchronization, etc will all make their way into the cache -- indeed, some of these already have."

The near-term outlook from Meta Group's Firstbrook? "Yet this year, we expect service providers like Akamai, telcos, and ISPs to offer managed ECDN services in the cloud. By 2004/05, we expect ECDNs to become much more closely integrated with both storage and computing infrastructure - for example, application servers."

An Established Player Speaks to the Future

Dr. Eric Brewer, cofounder and chief scientist of Inktomi (a company that, interestingly, just announced a partnerhsip with IBM in content-delivery technology), had some insightful comments on this topic at the "Storewidth 2002" conference for storage industry executives in April.

First, he described the basic CDN as "an overlay network to exploit edge resources, storage and CPU, to improve quality and delivery....Overlay networks give more control." But, he went on to comment on where he sees the Internet heading, and content delivery in particular:

"The 'old Internet' created widespread use of caches, which created CDNs. Caching is essentially storage in the network -- it was cheaper to store it than move it...Now, we're into a 'new Internet', with software to improve delivery of content. The old 'Net isn't going away, but this is a new layer on top of it. It involves redefining naming and routing, new services for fault tolerance, transforming data on the fly, and distributed control. It appeared almost silently, overnight."

Is a quiet revolution underway here? Stay tuned. It's bound to get noisier.