MEASUREMENT AND DESCRIPTION OF TIME-SHARING SYSTEM RESPONSE

The difficulties experienced in achieving precise definitions of commonly used response indices are discussed. A method for system evaluation is described which does not require precise definition of a response index. Particular attention is given to indices based on mean response time or response ratio, and the conditions under which either can be used. Alternative ways of counting users are discussed, and it is shown that different evaluations can require different types of count. Data from evaluations of two systems with contrasting workloads, one in a university and one in a commercial service bureau, are used to illustrate the points raised.