Yet the Internet is suffering from something more than just ‘growing pains’. Internet service providers and customers are often frustrated or constrained by the limitations of best-effort Internet. Neither the legacy networks transporting business nor the public Internet as it currently exists are economically viable for realizing advanced applications on a global, any-to-any basis.

The reasons are threefold. First, while the Internet Protocol (IP) provides ubiquitous connectivity it does not provide the security, reliability and quality of an essential communications infrastructure.

Second, today’s networks as service providers and vendors are repeatedly addressing the same problems with slightly different and incompatible methods.

Third, the IP business model has failed for service providers. Delivering higher and higher bit-rates over increasingly low-cost networks has become a path to financial instability and even insolvency.

The time has come, then, to combine the functionality and reliability of private networks with the ubiquity of the Internet.

What is envisaged is a public, packet-switched network based on IP and Multi-Protocol Label Switching (MPLS), and powerful enough to support all communications applications, securely and reliably, from anywhere to anywhere.

We call this the infranet.
The infranet model maximizes the value of the communications network while minimizing the replicated effort across the industry.

In particular, it aims to fully break the long held practice of building and operating a network for a single application. This is costly, complex and anachronistic. The infranet is intended explicitly to reflect the need to fully integrate computing intelligence with network services.

Infranets are able to support all the current network types, and in particular high-value services such as voice, video, business data and transactions, and gaming.

Infranets are purpose-built for supporting new and emerging services and new applications such as grid computing, and are flexible enough to manage dynamic shifts in networking requirements such as peer-to-peer traffic. Assurances for security, control, reliability and quality of service are delivered automatically for each application, enabling service providers to bill accordingly and profit from a diverse service mix.

For both consumer and business users this means that a far richer and more satisfying set of communications applications becomes possible.

Importantly, the infranet model also makes it possible to support these rich applications across the networks of multiple providers.

With this capability, service providers will want to carry traffic that originates on other networks as opposed to avoiding it. The bottom line for service providers is that the network can become a high-yield profit center as opposed to a low margin cost center.

Building such a universal, public network is the only economically viable alternative for the industry in the long run. MPLS is a cornerstone technology for infranets. Its emergence has enabled service providers to differentiate the handling of traffic based on application requirements. MPLS is also the first building block in the process of enabling applications to automatically reserve the resources they require through the network.

But MPLS and IP alone are not enough. Additional intelligence is essential to provide quality assurance, security, accounting and billing for new services. Greater control is also necessary to enable service providers to exchange traffic over their networks and be compen-sated for that carriage, just as they are today for carrying voice traffic.

Infranets can allow the creation of applications or content by adding a server resource as a network member, or by making available Web services whose applications are published in a service registry. This means that Infranets can accommodate both the legacy model of distributed computing and the new service-oriented architecture of Web services.

And because the IIC framework provides for interworking between services, infranets can also supply legacy services including voice, ATM, and frame relay.

Finally, the Infranet Inter-Carrier Interface (IICI) provides for deployment across multiple service providers around the world by connecting across the Signalling Stratum as well as the lower layers of the network. Specifically, the IICI defines accounting procedures for settlement and interconnection in a multi-carrier environment as well as controlling user identity across multiple service providers.

But the infranet initiative is not merely an exercise in defining standards and procedures. It is all about delivering services and ultimately revitalizing the business of the service providers themselves. For all the popularity and ubiquity of the Internet, it has not been a major profit centre for carriers. At a time when their traditional business is under threat from a number of fronts, telecom operators need a framework that will enable a successful IP business model. The goal for service providers is to capture higher-value applications in a way that rebuilds profit.

Right now the industry is in a transition period, with the arrival of low-cost mass-market data products, like MPLS IP-VPN, being arbitraged against high-cost legacy services like leased lines. A study by CIMI Corporation of the US service provider market makes the point that: “IP and the Internet have presented a model for mass-market data-driven public networking, and this model is the only one that can replace mass-market voice, whose revenue is now clearly tailing to zero.”

CIMI’s analysis finds that in the period 2004-2010, carrier revenues under the infranet model will be significantly higher over the non-infranet or limited infranet scenarios. But critically, the real growth is in content and applications delivery in the middle or transition years of the decade. Over 2006-07 period, ASP revenues are projected to be about 50 percent higher, and content revenues are up nearly 80 percent.

The difference is because service providers using the infranet model are better of to guarantee network performance and security for value-added applications. Content growth is even faster because the standardized environment offers fewer barriers to consumer adoption.

Similarly, the infranet model delivers gains when it comes to service provider profitability. Carrier internal rate of return (IRR) will rise from the current 14.5 percent to nearly 20 percent by 2010. For US carriers, that’s the difference between behaving financially as a utility, and being able to attract capital on par with other large-scale businesses.

Infranets can provide the business and services framework to reflate the profits of service providers in North America. But what of India?

India’s rapidly-growing communications industry faces different challenges. Provision of basic connectivity is a priority, and much of this is being delivered by wireless.

India has some 8 million Internet users today, but service providers are building some of the most extensive IP-based networks anywhere in the world. Nonetheless, the lesson from the more saturated North American and European data markets also applies here: high organic growth cannot prevent the erosion of margins and shareholder value by the current low value-added Internet business model. The primary value of the Internet is created not by the network, but by what is on the network – the e-commerce and content servers.

By providing a mechanism for integrating application and content features into the network, infranets offers telcos, in India and elsewhere, the opportunity to build a future business by selling higher-margin services, and not see profits wash out of the company by becoming raw bit-carriers.

Pradeep Sindhu, Vice Chairman and CTO founded Juniper Networks in 1996. He is now responsible for the company's technical roadmap and also plays an active role in design and development of future products. Dr. Sindhu holds a Ph.D. in Computer Science from Carnegie-Mellon University.