A STATISTICAL EVALUATION OF THE SCHEDULER OF AN EXPERIMENTAL INTERACTIVE COMPUTING SYSTEM**This research was partially supported by RADC contract AF 30602-70-C-0190.

Investigation into Scheduling for an Interactive Computing System

“…The model indicates that the dispatching of new inputs into service was also substantially affected by the level of congestion in the queues other than the foreground queue. It is worth noting that D had a significant effect on all the performance measures analyzed [5].…”

“…Only the characterization of system performance degradation in terms of the average value of reaction time will be discussed in this section. The same methodology was applied to characterizing both response time measures and is discussed in [5]. Reaction time is the elapsed time between receipt of an input and dispatching of the first time slice of service to the corresponding workspace.…”

“…which demand only interpreter servicing; and output-bound transactions, which demand interpreter servicing but generate output at such a high rate that their service frequently has to be interrupted to prevent them from being allocated all available output buffers. The different kinds of inputs generated by the active users during a ten-minute time interval were found to form a composite input traffic that was reasonably approximated by a Poisson process [5].…”