基于统计的软件系统自恢复时间阈值算法

Optimal Software Rejuvenation Schedule Algorithm under Statistical Minimum Cost

采用软件自愈技术来对抗软件的老化也会增加系统的成本。在此提出一种基于统计学的软件系统自恢复时间阈值计算算法。该算法基于满足软件系统自恢复系统成本开销最小化的思想,通过TTT变换推导得出;可以在系统性能衰退时间分布未知的情况下,根据一定量的性能衰退检测数据,计算出优化的软件系统自恢复时间阈值。仿真实验结果表明:计算结果合理、稳定性好,能有效地应用于实际系统中。

The execution of software rejuvenation to counteract the software aging phenomenon can involve an overhead. This paper develops a statistical algorithm to estimate the optimal software rejuvenation schedules, provided that the statistical complete sample data of failure times is given, and the optimal software rejuvenation schedules which minimize the expected rejuvenation cost in per unit time are derived under statistical TTT transitions. Numerical results illustrate the resulting estimators for the optimal software rejuvenation schedules have quite nice convergence properties and are useful in applying to a real software in operation without specifying the underlying failure time distributions.