Analytic Models of Adaptive Load Sharing Schemes in Distributed Real-Time Systems

  • Authors:
  • K. G. Shin;C. J. Hou

  • Affiliations:
  • -;-

  • Venue:
  • IEEE Transactions on Parallel and Distributed Systems
  • Year:
  • 1993

Quantified Score

Hi-index 0.00

Visualization

Abstract

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 anoverloaded node may miss their deadlines even if the overall system has the capacity tomeet the deadlines of all tasks. A decentralized, dynamic load sharing (LS) scheme hasbeen proposed as a solution to this problem. Analytic queuing models to comparativelyevaluate this LS scheme as well as three other schemes-no LS, LS with random selectionof a receiver node, and LS with perfect information- are developed. The evolution of anode's load state is modeled as a continuous-time semi-Markov process, wherecumulative execution time (CET), rather than the commonly-used queue length (QL), isemployed to describe the workload of a node. The proposed scheme is compared againstother LS schemes. The validity of analytic models is checked with simulations. Bothanalytic and simulation results indicate that by using judicious exchange/use of stateinformation and Bayesian decision mechanism, the proposed scheme makes a significantimprovement over other existing LS schemes in minimizing the probability of dynamicfailure.