A choices implementation of the Universal Scheduling System

A multiprocessor system often has more jobs ready to execute than processors available on which to run them. In such cases, some form of process scheduling is needed to determine which processes may run on which processors. This scheduling activity is similar to that of a single processor system except that in addition to a process, a processor has to be selected. There are many different scheduling policies to determine which process will execute next. The choice of the policy depends on the goals of the applications in the system. Different applications might need different policies or a single application might use different policies for different subsets of the processors. The Universal Scheduling System (USS) described in [1][2] classifies scheduling policies according to a simple abstract model. A scheduling algorithm is specified in terms of a deci­sion mode, a priority function, and an arbitration rule. This model makes it easy to define many different scheduling algorithms and provides a simple, efficient, and easily expandable framework for their implementations. This thesis uses the Universal Scheduling System as a model for an object-oriented implementation of schedulers in the Choices [3][4][5] operating system.U of I OnlyRestricted to the U of I community indenfinitely during batch ingest of legacy ETDs