A class of scheduling policies for real-time processors with switching system applications

Consider a queueing system where a single server has to process a queue of primary jobs and also perform background tasks. In this paper, we analyze a class of scheduling policies that allot an uninterrupted segment to background work whenever no primary jobs are present and also when a limit on primary job processing is exceeded. The following performance measures are derived: the delay experienced by primary jobs; the degree of protection given to background work; and the overhead incurred. The results are then used to determine the scheduling parameters that satisfy several performance objectives.