The Optical Internet: A Service Enrichment Strategy

Telecommunications, Nov, 2000 by Alex Dobrushin

The real challenge for service providers is how to generate more revenues from their IP/optical backbones.

IP traffic will represent more than 90 percent of the total public network traffic by 2002, according to analyst group RHK. Not only is IP set to dominate, but Internet traffic will expand more than tenfold over the next three years, reaching more than 100,000 Gbps in 2003. To accommodate this deluge, service providers plan to invest more than $70 billion in core routing and optical transmission equipment to significantly expand their IP/optical backbone networks. Yet revenues from IP services will only approach $25 billion, just 30 percent of the total network services revenue pool of $75 billion, according to RHK.

One key cause of the IP revenue dilemma is competition in the Internet access market, which has lead to commodity; flat-rate pricing. The most popular way businesses access the Internet is over T1. Extensive use of graphics, audio and video content has driven average utilization up significantly, yet the user is charged the same rate. Service providers must add capacity in the network core without any corresponding increase in revenue. The real challenge for service providers is how to generate more revenue from their IP/optical backbones.

The solution lies in mastering service delivery over IP. By taking advantage of the latest advances in IP QoS, MPLS (multiprotocol label switching), and service transformation technology (the conversion of non-IP services to IP), service providers can begin to evolve dedicated IP infrastructures into a new network architecture called the optical Internet. An alternative to operating separate service-specific networks, as is customary today, the optical Internet is a single multiservice network using IP as the underlying protocol for all service delivery. It allows service providers to supplement their IP revenues with other established network service revenues from frame relay, TDM private lines and ATM and address an additional $50 billion in service revenue that will be available in 2003. Adopting service delivery over IP results in faster payback of the tremendous carrier investment in IP/optical networks (see Figure 1).

Getting Down to Basics

Virtually all communication carriers are embracing the optical Internet. Backbone operators, for example, see IP based transport as way of streamlining and simplifying their operations. ILECs and IXCs view it as a solution to scalability issues that trouble their most popular services. For instance, frame relay service delivered over a network populated with frame relay and ATM switches requires carriers to set-up N x (N-1) number of PVCs (permanent virtual circuits) for every N sites served. This is known as an "N-Squared" problem. Delivering frame relay over IP reduces the number of PVCs that must be maintained to just one per site. Next-gen core IP routing switches also scale to greater capacity (e.g., more OC-48 and OC-192 ports) than ATM switches under development, with Tbps-class switches on the way.

This new network architecture also gives carriers a clear migration path from running multiple service-specific overlay networks to a converged network based on plentiful bandwidth and IP that is capable of delivering multiple services. In the new competitive environment, the "If you build it, they will come" approach to implementing overlay networks before the market demand is known is not wise or economical. The multiservice nature of the optical Internet avoids this guessing game because it is a common infrastructure from which services can be turned up based on local demand. This makes out-of-region expansion, for example, more cost effective and less risky.

Service providers are raking advantage of IP QoS and MPLS traffic engineering to capitalize on the low operating cost, high bandwidth, and better flexibility of the IP/optical environment. For the optical Internet to evolve into a multiservice, converged network, carriers need to address two additional key requirements:

* Fault tolerance to ensure that services maintain a level of availability equivalent to or better than what exists in today's overlay networks. In general, carriers need to achieve 99.999 percent or better availability.

* Service transformation to provide the adaptation of non-IP services for IP/MPLS transport, while maintaining the integrity of service-specific characteristics such as alarms, timing and transparency to the subscriber.

Overlay networks are generally connection-oriented Layer 2 networks with Layer 3 services running over them. It is fairly easy to establish fault tolerance in these networks. With all the services running over a Layer 3 protocol, the optical Internet presents a new challenge. Network resiliency must be elevated to Layer 3.

Fault tolerance must start at the network edge where services converge. A new breed of routers at the service edge, called ASRs (aggregation service routers), are capable of restoring services with SONET-like speed at less than 50 ms to eliminate a point of failure (see Figure 2). To accomplish this, ASRs support a combination of hardware and software that provide stateful redundancy of routing states and tables. This capability is essential for recovery from a primary controller failure without affecting the services delivered. A standby controller with stateful IP routing sessions, such as BGP (Border Gateway Protocol), OSPF (Open Shortest Path First), and packet forwarding information steps in and restores the services being delivered.