An IPSS-Based Model-Building Methodology for Ranking and Evaluating Computer Hardware/Software Systems

摘要

This research sought to design and implement a model building methodology for simulating US Army computer hardware/software systems. Computer systems are characterized in terms of file parameters, hardware specification, and software use of files. These descriptions reside in a model library and are the building blocks in the model synthesis process. The Information Processing System Simulator (IPSS) language was used to encode these descriptions and to represent the sequence of computer activities for application program processing (e.g., job scheduling, buffer management, channel program). Two computer systems were compared using this method. Simulation models were written for an IBM 360 Model 30 computer and a Honeywell Level 6 minicomputer. A subset of the US Army Standard Installation/Division Personnel System (SIDPERS) provided a common loading for both systems. Data was collected on an operational IBM 360/30 and the IPSS model was validated. Statistical results, derived from IPSS, indicate resource utilization (for both hardware and software resources), elapsed time, and queueing. Our results project that an 8-hour execution of SIDPERS on the IBM 360/30 would execute in approximately 2-1/2 hours on the Honeywell machine. Models of several hardware variations were prepared in order to demonstrate responsiveness capabilities of the method. A manpower analysis is provided for guidance in estimating future work.