Akamai Technologies, Speedera Networks and other CDSPs use "overlay" networks of distributed servers with primary and secondary storage that mirrors content on original servers owned by subscribers. The CDSP provides a customer with the tools necessary to convert or redirect URLs from within that customer's own site to edge devices that mirror original content close to users from an ISP's access network or an enterprise data center. Edge devices balance loads and act as buffers when heavy demands overtax a content owner's server resources and bandwidth capacity. They can also act as multicast splitters for live streaming events (for more on this technology, see "Tutorial: The Wizardry of Multicast").
The CDSP market, like other high-technology sectors, has suffered from the effects of recent dot-com failures and spending cuts. But reducing infrastructure costs and maximizing content-retrieval time have a significant impact on a content owner's bottom line. This market segment will grow to $703 million by 2005, the Aberdeen Group projects. And Meta Group estimates that more than 14,000 enterprises will use CDSP services.
With this growth in sight, hosting providers, ISPs and large telecommunication providers -- including AT&T and Qwest Communications -- now compete with established CDSPs for enterprise customers.
Whether CDSP services come from a hosting provider, an ISP or from a traditional CDSP will depend on how well the providers enhance and expand their service offerings. Akamai says it hopes to deliver custom advertising insertions and content personalization based on bandwidth, location and service level with ESI (Edge Side Includes). Digital Island says it is looking to leverage its "on-demand overlay routing" algorithm to discover optimal network paths for content delivery dynamically.
Digital Island also has assumed a major role in content-peering initiatives to foster interoperability among CDSPs and hosting and telecommunication providers. Such initiatives aim to let the providers share networks, content and revenues, thereby expanding the market for content-networking services. Content Bridge Alliance and Content Alliance are two such efforts for content peering. Although Digital Island is a company to watch in 2002, the alliances require members to align complex business processes to maintain service levels among members. It is unlikely these issues will be resolved in 2002.
Customers pay for the amount of storage used on CDSP edge devices and the amount of traffic their content generates in bytes. Under this business model, a customer accrues fairly low switching costs to mirror its data on another provider's edge device and to redirect URL requests to alternative content-delivery solutions.
For enterprises with Internet content to publish in 2002, Akamai and Digital Island have an edge on the competition. But the catchphrase to remember when it comes to any service with low switching costs is: Keep your contracts short and your options long.
The same content-delivery technologies CDSPs provide on the Internet can reduce latency and speed content to end users on enterprise networks. Corporate communications, education, collaboration solutions and similar content can be delivered via overlay networks using videoconferencing, Web casting and Web conferencing applications. These overlay networks include server and network caching devices, switching and routing techniques, and end-to-end content-management solutions.
The CDN service model has proved its mettle for delivering content over the Internet, reducing the high-latency effects of bottlenecks and congestion on public networks. It remains to be seen how broadly this model will appeal to enterprise networks that have better control and management over their own resources and have substantial investments in infrastructure and storage.
ECDN providers need to go beyond the edge and provide interoperability with existing infrastructure products, storage solutions and applications. Ideally, any new solution will work with existing cache stores and provide application integration as well as address security and digital-rights management.
Delivering enterprise content is inherently different from delivering Internet content. Enterprises have captive audiences and exclusive rights of way to end users. ECDN providers will have to show clear paths of integration with existing infrastructure and ROI for end-to-end content management solutions. Otherwise, enterprises will continue business as usual in content delivery in the coming year.
MSPs: Believe the Hype?
Look hard at the ubiquitous XSP acronym. The X isn't really an X at all. Look harder, and you'll see it's really the twisted, mangled blades of the great PR hype fan still wobbling along in the wind. In the case of managed service providers, this condition is particularly acute.
Most of the players in the MSP space have no one to blame but themselves. At the height of the MSP hype, many of these companies threw talented but unfocused resources at every sort of network-management problem. The winners in this space are going to be those that are in it for the long haul and have the operational chops to really run network operations.
But despite the gloomy economy, the MSP market is expected to take off. "The management service provider market will conservatively reach $10 billion in revenue by 2005, attracting many vendors," says Corey Ferengul, at Meta Group, in Stamford, Conn. "We expect (and have seen) vendors from diverse backgrounds to enter the MSP arena: independent software vendors, systems integrators, Internet service providers and traditional outsourcers." But in the short term, look for many MSPs to do an exodus (pun intended).
In 2002, we'll see a major shakeout in this space. Many hosting and infrastructure service providers have tried to hang a MSP shingle on their infrastructure and raised-floor offerings. If you decide to hire an MSP, make sure the company can run its service in a production environment. There's no doubt that established players, including Electronic Data Systems (EDS) and IBM, want a slice of the MSP pie. At this point, however, the only sweet spot for MSP customers are medium-sized enterprises with new (read: Web) applications and Web infrastructure. And companies like IBM and EDS aren't well-positioned to meet this market. Some smaller, new companies, such as Nuclio and SiteLite, may be.
For most companies, outsourcing the NOC is a hard sell. The people who staff the NOC are valuable. They make the trains run on time, and there's generally no reason to mess with them. Also, outsourcing is easier to set up for a new technology, than it is for something that reaches deeply into the internals of a business and probably back to the beginning of time.
It is a good bet that as businesses analyze their own core competencies, NOC and operations won't have a compelling revenue story to tell. And at that point, smart CIOs are going to look long and hard at the organizations with which they partner for this mundane but critical function.
Darrin Woods is a technology editor of Network Computing. Before joining the magazine, he worked as a WAN engineer for a telecom carrier. Send your comments on this article to him at dwoods@nwc.com.
Bruce Boardman, Sean Doherty, Lori MacVittie, Steven J. Schuchart Jr. and Greg Shipley contributed to this article.
REPORTS
ANALYTICS
Follow 
