Computer Systems Analysis Methodology: Studies in Measuring, Evaluating, and Simulating Computer Systems.
A briefing on tools and techniques developed in four research efforts in the analysis of complex computer systems (R-513, R-573, R-584, RM-6132). (1) Computer system characteristics and activity can be conveniently modeled in the prototype ECSS (Extendable Computer Systems Simulator) language while maintaining the flexibility and power of SIMSCRIPT II. Standard aspects are handled by service routines, and the user can insert his own ECSS or SIMSCRIPT routines at will. (2) CPU utilization and throughput can be improved 30 to 300 percent by "tuning" multiprogrammed systems. Small changes in workload characteristics can produce large changes in performance. (3) Choosing computers by their performance on benchmark jobs can be disastrous unless the benchmarks are representative of the workload. (4) The complexity of multiprogrammed systems requires techniques for isolating key factors in any effect. (5) In controlled experiments in interactive man-computer problemsolving, time constraints on console use produced dissatisfaction--but also produced better results and more efficient work. 53 pp. Ref.