ISSRE 2006 - Using Stochastic Models in the Testing Process

Using Stochastic Models in the Testing Process

Robert Musson

The 17th IEEE International Symposium on Software Reliability Engineering (ISSRE 2006) -- Industry Practices (ISSRE 2006)
Raleigh, North Carolina, USA, 6-10 November 2006

Abstract

Stochastic modeling is a technique that uses probability equations to simulate real life systems. The Microsoft Windows team has been using stochastic models with great success to manage the test effort. By modeling the interactions that occur between development and test, the test team has been able to determine organization capabilities and use them to predict the outcome of the test effort.

A number of interesting results have been seen in the modeling effort. First, there is an optimal rate of feature delivery beyond which the throughput rate will actually decrease. Second, improving the efficiency of test is counter-productive if the entire organization is not improved. There is actually the possibility of too much testing. Finally, even simple models have been seen to be highly predictive for quality and schedule. By closely monitoring the predicted versus actual trends in defect discovery, the team has been able to make changes to the test processes.

START Conference Manager (V2.52.6)

Maintainer: mark.sherriff@ncsu.edu