We characterize load sharing with three component policies: the transfer policy, the location policy, and the information policy, and carefully tailor each policy to enable ``capable'' nodes to share the loads of ``incapable'' ones and to maximize the probability of aperiodic tasks meeting their deadlines.

• We appropriately incorporate task laxity into the transfer policy, where task laxity is the latest time a task must start execution in order to meet its deadline.
  
  
• With the use of Bayesian analysis, we minimize the probability of transferring an overflow task to an `incapable node'' due to the use of out-of-date state information.
  
  
• With the use of priority lists, we minimize the probability of multiple nodes sending their overflow tasks to the same ``capable'' node.
  
  
• With the use of queuing theory analysis, we avoid excessive task transfers by minimizing the probability that a task transferred to a remote server node has to be transferred out again due to the arrival of tighter-laxity tasks at the remote node.
  
  
• With the use of parameter estimation, we devise a timeout mechanism with on-line adjustable timeout periods to deal with the problem of falsely diagnosing a silent but healthy node as faulty.

The research results have been reported in the following papers:

• Kang G. Shin and Chao-Ju Hou, "Analytic models of adaptive load sharing schemes in distributed real-time systems," IEEE Trans. on Parallel and Distributed Systems, Vol.4, No.7, pp. 740-761, July 1993.
  
  
• Chao-Ju Hou and Kang G. Shin, "Incorporation of optimal timeouts into distributed real-time load sharing," IEEE Trans. on Computers, Vol.43, No.5, pp. 528-547, May 1994.
  
  
• Kang G. Shin and Chao-Ju Hou, "Design and evaluation of effective load sharing in distributed real-time systems," IEEE Trans. on Parallel and Distributed Systems, Vol.5, No.7, pp. 704-719, July 1994.
  
  
• Chao-Ju Hou and Kang G. Shin, "Load sharing with consideration of future task arrivals in heterogeneous distributed real-time systems," IEEE Trans. on Computers, Vol.44, No.9, pp. 1076-1090, September 1994.
  
  
• Kang G. Shin and Chao-Ju Hou, "Evaluation of load sharing in HARTS with consideration of its communication activities," ACM 1993 SIGMETRICS Conf. on Measurement and Modeling of Computer Systems, extended abstract pp. 270-271, Santa Clara, California, May 10-14, 1993. An enhanced version has been submitted to IEEE Trans. on Computers.
  

---

Return to Project Home Page
Date last modified -- August 15, 1998
Direct comments concerning this WWW site to: jhou@ece.osu.edu