Estimation of Run Times using Signature Table Analysis.

Abstract
Algorithms for managing jobstreams in a complex computer environment often rely on various estimates of job run times. Due to wide variability of run times from one execution of a job to another, point estimations of run times are fairly unreliable. An alternate approach to using point estimations is to use intervals which span the range of possible run time values. In an interval approach run times can be predicted with respect to membership in one of a limited set of run time intervals, with relatively high confidence. This paper presents a formal methodology for run time estimation based on an interval approach. The estimation is done using signature table analysis and is accompanied by a statement of statistical confidence in the results. (Author)

This publication has 0 references indexed in Scilit: