Optimal Rejuvenation Scheduling of Distributed Computation Based on Dynamic Programming
Hiroyuki Okamura, Kazuki Iwamoto, Tadashi Dohi · Journal of Computer Science · 2006
Recently, a complementary approach to handle transient software failures, called software rejuvenation, is becoming popular as a proactive fault management technique in operational software systems. In this study, we develop the optimal scheduling algorithms to trigger software rejuvenation in distributed computation circumstance. In particular, we focus on two different computation circumstances in terms of detection of failures. Based on the dynamic programming, we derive the optimal software rejuvenation schedule which minimizes the expected total time of computation. In numerical examples, we examine the sensitivity of model parameters characterizing the failure phenomenon to the resulting optimal rejuvenation schedule.