摘要
该文针对软件在连续运行过程中普遍发生的老化现象,提出了一种嵌套的基于时间的软件抗衰策略,并采用Petri网来描述每次恢复的具体过程,采用有限状态自动机控制恢复次数,建立整个策略的实施过程模型。与传统的基于时间的软件抗衰策略不同,该策略在每次执行系统级恢复前首先执行N次的应用级恢复,从而减少了恢复时间,降低了抗衰成本和周期性应用级恢复策略预测失败的风险,提高了系统的可靠性。另外,该策略具有可扩展性,对于多种因素导致的性能衰退,可以执行多级嵌套的恢复。
Software, aging is a very common phenomenon for the large server system during a long time of running. The simple periodic software rejuvenation results in more spending. A new nested time-based software rejuvenation policy is put forward. In order to implement this policy efficiently, the software working process is modelled by Petri net and the rejuvenation times are controlled by finite-state automata. Comparing with the current rejuvenation policies, the nested time-based software rejuvenation policy executes firstly some times of application-level rejuvenation before one time system-level rejuvenation, which makes it possible to cost less time, to reduce the risk of failure and to enhance the software reliability and availability. In addition, this policy is also extensible. The multilevel nested rejuvenation can be carried into execution if the performance degradation is induced by several factors.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2006年第1期43-47,共5页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金(60273035)