Worst-Case Analysis for Real-Time Processes in Time-Sharing Environments

Abstract

This paper presents a technique for predicting response times of real-time tasks that operate in a time-sharing environment that also has priority classes (as in the real-time priority class paradigm of the Solaris operating system). Response time prediction is accomplished by a worst-case approach for calculating the queuing delay that a process will experience. Experimental results show that the predicted response times for actual programs provide feasible upper bounds, making the prediction technique suitable for hard real-time environments.