# Sensitivity Study of the Clustering Approach to Workload Modeling

### Maria Calzarossa and Domenico Ferrari

###
EECS Department

University of California, Berkeley

Technical Report No. UCB/CSD-85-222

January 1985

### http://www.eecs.berkeley.edu/Pubs/TechRpts/1985/CSD-85-222.pdf

If an adequate separable model of an interactive system can be built, if the users' behaviors can be accurately modeled by probabilistic graphs, and if the workload and the model of it to be constructed are stationary, then a perfectly accurate workload model of a given workload can be constructed by grouping together all commands with identical characteristics. If a clustering technique which groups together commands with similar characteristics is applied, is the workload model produced still acceptably accurate? This paper addresses this question (and other similar ones) by taking an experimental approach. The answer is that, in the case considered here, the accuracy is still acceptable if certain conditions are satisfied.

BibTeX citation:

@techreport{Calzarossa:CSD-85-222, Author = {Calzarossa, Maria and Ferrari, Domenico}, Title = {Sensitivity Study of the Clustering Approach to Workload Modeling}, Institution = {EECS Department, University of California, Berkeley}, Year = {1985}, Month = {Jan}, URL = {http://www.eecs.berkeley.edu/Pubs/TechRpts/1985/5925.html}, Number = {UCB/CSD-85-222}, Abstract = {If an adequate separable model of an interactive system can be built, if the users' behaviors can be accurately modeled by probabilistic graphs, and if the workload and the model of it to be constructed are stationary, then a perfectly accurate workload model of a given workload can be constructed by grouping together all commands with identical characteristics. If a clustering technique which groups together commands with similar characteristics is applied, is the workload model produced still acceptably accurate? This paper addresses this question (and other similar ones) by taking an experimental approach. The answer is that, in the case considered here, the accuracy is still acceptable if certain conditions are satisfied.} }

EndNote citation:

%0 Report %A Calzarossa, Maria %A Ferrari, Domenico %T Sensitivity Study of the Clustering Approach to Workload Modeling %I EECS Department, University of California, Berkeley %D 1985 %@ UCB/CSD-85-222 %U http://www.eecs.berkeley.edu/Pubs/TechRpts/1985/5925.html %F Calzarossa:CSD-85-222