Identifying Service Requirements

If the NSP is building a vertical system, then there is a lot of flexibility to make tradeoffs between client capabilities and server capabilities. Information can be staged to capable clients in time for its use by location-aware asynchronous delivery. If, for example, the application is shuttle scheduling and monitoring on a corporate campus, dynamic conditions such as measured recent travel time on specific links within the campus road network can be delivered on a communications-available basis. When a new building goes online or a road is opened or closed, the information can be scheduled for delivery to each shuttle before the new information would have any bearing on travel.

Peak, average, and other measures of request volume are needed in order to scale the services. A typical core navigation server can handle a few transactions per second with latency (round trip delay) of a few seconds. Relaxing the peak performance requirements can significantly reduce system cost. Unfortunately system behavior is often judged by its worst-case (i.e. peak load) performance.

Previous Next