Introduction to statistical pattern recognition (2nd ed.)
Introduction to statistical pattern recognition (2nd ed.)
On the construction of a representative synthetic workload
Communications of the ACM
Software unit profiles & Kiviat figures
ACM SIGMETRICS Performance Evaluation Review
A critical overview of computer performance evaluation
ICSE '76 Proceedings of the 2nd international conference on Software engineering
On state-dependent workload characterization by software resources
SIGMETRICS '82 Proceedings of the 1982 ACM SIGMETRICS conference on Measurement and modeling of computer systems
Capacity planning for MVS computer systems
ACM SIGMETRICS Performance Evaluation Review
Finding representative workloads for computer system design
Finding representative workloads for computer system design
Hi-index | 0.00 |
A major problem in the evaluation of the performance of a multiprogrammed, time-sharing computer system is the development of an accurate description of its normal workload. In evaluating the performance of a computer system, the validity of the analyzed data can be no better than the degree to which the test workload represents the real workload. The paper reports on the research aimed at developing a probabilistic model of the workload which can be validated with respect to the real workload and is easy to use in a performance evaluation study. In this characterization, the workload is modelled as a mixture of a number of classes. The distribution of load in each class is described by a unimodal distribution of various resource loadings. Therefore, the workload is modelled as a mixture distribution in an appropriately chosen n-dimensional space. By increasing the number of classes the mixture distribution model allows us to model a complex multimodal distribution. At present, we are studying the workloads on the Univac 1108/1106 computer at the Computer Science Center of the University of Maryland. This installation is used for educational and research computing by the students and faculty at the University and handles a workload of several thousand processes during a typical day. Some results of this study are presented.