Methods of software performance monitoring and data reduction, recently implemented at the University of Wisconsin, have been successfully used to provide a meaningful set of information for management decision making. Emphasis in performance measurement has been placed on I/O request data, resource utilization, user and system file usage summaries, and individual job accounting statistics.The major goals at Wiscon in have been to maximize system through put and to reduce turnaround time for small jobs, while keeping turnaround for larger jobs within a reasonable limit. Over a two year period throughput was increased by a factor of two and mean turnaround time was decreased by a factor of ten.Although implemented for a specific system, the UNIVAC 1108, the overall approach is applicable to a wide range of general purpose multiprogramming time sharing systems. The description of the data collection and analysis facility emphasizes this generality.
Unfortunately, ACM prohibits us from displaying non-influential references for this paper.
To see the full reference list, please visit http://dl.acm.org/citation.cfm?id=1146439.