Two definitions of efficiency

In focusing on user preferences, we need to distinguish two very different notions of efficiency:

  • Network efficiency refers to the utilization of network resources, such as bandwidth and buffer space.

  • Economic efficiency refers to the relative valuations users attach to their network service.

If a network can maintain a target level of service while minimizing the resources needed to provide this service, we say that its operation is network efficient. For example, by statistically multiplexing bursty transmissions, the bandwidth required can often be reduced from that of a pure circuit-switched approach while still meeting the delay requirements of the applications.

If no user currently receiving a particular QOS values it less than another user who is being denied that QOS, we say that operation is economically efficient. For example, if one user is willing to pay x per second for undelayed access to a 1 Mbps link, and a second user is willing to pay only X/2, and if only one of them can be accommodated, then in an economically efficient network the bandwidth will be allocated to the first user (whether or not they actually pay anything).

An obvious question is, why will either type of efficiency continue to be important? Some observers have suggested that the widespread deployment of fiber optic lines, and continuing exponential decreases in processor and memory costs, will result in these network resources becoming essentially ''free'' so that efficiency in their use will not be important in the future, and all users can always be accommodated. We do not believe these arguments apply in the short or medium terms, if indeed they will ever apply. User demands are increasing exponentially, so that it is not clear when—if ever—network resources will be ''free''. We share the dream of ubiquitous, two-way broadband connectivity at low or zero cost, but believe we must wait at least a few decades to achieve it, despite astonishing technological advances. Consider the cost of providing gigabit bandwidth not just to every home in the industrialized world, but for the other three-quarters of the planet's population as well. Add to that the cost of gigabit mobile communication that will follow each person around town, country and world. And experience suggests that application developers will have little difficulty in designing new services that use up all the available resources.

For the foreseeable future, we will continue to live in a world characterized by network resource scarcity. We will move most quickly towards ''free'' service if we use our scarce network resources — whether public or private — efficiently in economic terms.

Meanwhile, if commercial providers are not responsive to user valuations, they will not succeed in a competitive market. The same considerations apply even to private-access networks: the ultimate goal is to maximize some human measure of the value of using the network, such as profits, sales, shareholder value, and so on.