IP routing key to integrated services

For years, people in the industry have talked about providing fully integrated voice, video and data services through the same network equipment. Few fully integrated services actually exist, however. Because voice networks were once the only game in town, data services had been integrated onto the voice backbone for a period of time, accelerated by the breakup of the AT&T monopoly, which forced the Baby Bells to sell bandwidth to competitive carriers and to customers. Eventually, the voice backbone could not handle the data traffic and, in recent years, the Internet and bypass solutions have separated that traffic from the voice backbone. In essence, integration had become dis-integration.

During the past decade, the industry viewed asynchronous transfer mode (ATM) as the network infrastructure that would unite all services and bring quality-of-service (QoS) to the network. Although carriers implemented ATM in their network cores and as a backbone for frame relay services, they found ATM to be too expensive to move it fully to the edge and access of the network. The cost-efficiencies, speed and technical advantages of Ethernet, Internet Protocol and routed services outmatched ATM.

New standards

Early on, carriers promoted new standards, like ATM, that they believed would become ubiquitous. Instead, Ethernet, IP and packet networks became universal, even though they lacked the underlying QoS features that would enable integrated services. The question facing the industry now is, how will QoS features be integrated on top of an unreliable network infrastructure?

The Internet Engineering Task Force has worked on answering this question. Collaborative efforts produced a number of projects that, when put together, will provide the basis for integrated-services initiatives. Researchers based the core of their work on the multiprotocol label switching (MPLS) standard and associated protocols. These researchers found that providing IP-based QoS called for an entirely new paradigm.

Routing traffic

The ATM pioneers helped developers determine how to map routed traffic to switching networks, but something was missing: a simple way to switch traffic in the network rather than relying on complex routing protocols, such as BGP, which are difficult to manage and slow packet delivery. The result of the research was MPLS (RFC 3031), which defines a simple switching paradigm using labels to switch traffic at high speeds.

MPLS provides a way for the network core to switch packets at high rates without the cost and delays associated with routing protocols. MPLS is a good start for the core, but more enhancements, such as signaling protocols, are needed.

The ability to deliver QoS features in the network seems to depend on the end devices' support of the IP. However, what if the end devices are palmtop systems or cellular phones? There are those who believe that IPv4, the current version of the Internet Protocol, is simply not up to the task of supporting palmtops and cellular devices. Their doubts stem from a shortage of Internet addresses, plus a lack of integrated features. For the next generation, IPv6, implementations have been IPv6 native, end to end. Most IPv6 implementations will be on end equipment and access devices, the areas that really need more IP addresses.