Empirical methods for estimating workload capacity

Abstract

We explore experimental procedures for comparing the capabilities of complex discrete event service systems. Instead of measuring system capability by analyzing or simulating the system with a constant rate of arriving work, system capability is measured as the maximum rate of work arrival for which the system has a steady state. Hence, we seek the arrival rate which causes the system to be at full capacity. This rate is arguably the best indication of the service system's capability. We treat both work-conserving and non-work- conserving service systems, using traditional and specialized measures of system performance