Evaluating system effectiveness in high performance computing systems
Adrian T. Wong · eScholarship (California Digital Library) · 1999
High performance scientific computer systems traditionally have been compared using individual job performance metrics.However, such metrics tend to ignore high-level system issues, such as how effectively a system can schedule and manage a varied workload, how rapidly the system can launch jobs, and how quickly it can recover from a scheduled or unscheduled system outage.Yet high-level issues such as these are now among the most important factors in the overall effective performance of a system.This article discusses these issues in some detail and then describes a new performance test, called the "Effective System Performance (ESP)" benchmark, which we are developing to measure system-level performance.