Analytic models of adaptive load sharing schemes in distributed real-time systems

In a distributed real-time system, nonuniform task arrivals may temporarily overload some nodes while leaving some other nodes idle. As a result, some of the tasks on an overloaded node may miss their deadlines even if the overall system has the capacity to meet the deadlines of all tasks. A decentralized, dynamic load sharing (LS) scheme has been proposed as a solution to this problem. Analytic queuing models to comparatively evaluate this LS scheme as well as three other schemes-no LS, LS with random selection of a receiver node, and LS with perfect information- are developed. The evolution of a node´s load state is modeled as a continuous-time semi-Markov process, where cumulative execution time (CET), rather than the commonly-used queue length (QL), is employed to describe the workload of a node. The proposed scheme is compared against other LS schemes. The validity of analytic models is checked with simulations. Both analytic and simulation results indicate that by using judicious exchange/use of state information and Bayesian decision mechanism, the proposed scheme makes a significant improvement over other existing LS schemes in minimizing the probability of dynamic failure