A Distributed Scheduling Algorithm for Heterogeneous Real-Time Systems

Much of the previous work on load balancing and scheduling in distributed environments was concerned with homogeneous systems and homogeneous loads. Several of the results indicate that random policies are as effective as other more complex load allocation policies. In this paper, we investigate the effects of heterogeneity on scheduling algorithms for hard-real time systems. We propose a distributed scheduler specifically to handle heterogeneities in both nodes and node traffic. The performance of this algorithm is measured in terms of the percentage of jobs discarded. While a random task allocation is very sensitive to heterogeneities, our algorithm is shown to be robust to such non-uniformities in system components and load.