This is similar to the premium service that the postal service offers to its customers called express delivery. Dan only pays a small fraction of what he would need to pay for a higher bandwidth access since this would only be for a short period of time and only for certain traffic. The telco would provide an instantaneous high-capacity link through its network between the Netflix movie server and the user’s PC. The online movie service provider charges the user through its existing billing relationship and shares the extra revenue with the telco. The online service provider can guarantee its customers a great experience, and the telco is able to monetize its expensive networks that it needs to upgrade and maintain. What is more, this scenario is net neutral to the customer, since he still has a “best effort” network, but can choose a higher bandwidth for a certain service on demand. Still, this partnering scenario between the online service provider and the telco generates a number of questions with regard to technical, regulatory, and business aspects.

Technical Feasibility of On-Demand Bandwidth and Ad-Hoc QoS

For the partnering scheme to be successful, the telco needs a technical infrastructure that provides on demand guaranteed quality of service on the whole network path between the service provider’s servers and the customer’s computing equipment. This end-to-end view of network quality covers the telco’s backbone network (core, edge, distribution), any aggregation network that might exist, and the access network into the customer’s premises.

Least problematic is the telco’s backbone network, since the IP/MPLS technology often deployed can handle on demand QoS requests and traffic engineering through a flexible label switching mechanism. Packet streams can be differentiated by service and forwarded easily through the network. Nor do these fast core networks suffer from overbooking.   A different situation exists in the currently deployed aggregation networks, which usually have high overbooking factors and no consistent quality of service schemes. An ATM-based DSL aggregation network needs to be upgraded or replaced by Ethernet-based aggregation which provides both QoS and bandwidth on demand and superior OAM (operations, administration, and management) capabilities. This can be a major cost factor for network transformation.

Among all currently installed access technologies, the various forms of DSL have been most widely deployed by ISPs worldwide. On the upside, the DSL Forum has defined mechanisms for differentiating traffic streams. The DSL specifications provide a “turbo button” feature which can give on demand high bandwidth to a subscriber. This enables ISPs to provide “best effort”, i.e., cheap connections, to all users while providing the possibility to get on-demand bandwidth and QoS. This is compatible with ATM as well as Ethernet DSL aggregation networks (TR-59, 101). On the downside, these mechanisms might not be enabled in many current network implementations and would need to be provided through costly upgrades.