基于检查点和Rejuvenation的软件运行模型分析

Analysis of A Software Operation Model on Checkpointing and Rejuvenation

近来的研究表明 ,长时间运行的通信软件往往存在老化现象。为防止软件老化引起的突发性系统停机 ,提高系统的可靠性 ,人们提出了一种预防性的软件容错策略 ,称为rejuvenation。由于它的过程复杂 ,总的停机成本仍然是可观的。检查点是一种轻量级的软件容错策略 ,其成本远小于rejuvenation的成本。该文通过合理结合rejuvenation和检查点技术 ,实现了降低总的系统停机成本的目的。文中给出了系统的Petri网模型 ,并结合实例进行了分析。

Recent study shows that the phenomenon of software aging always exists in long-running communication software. In order to prevent the unplanned system failure due to aging and improve system reliability, software rejuvenation, a preventive fault-tolerance approach is proposed. Because of its complex procedures, the downtime cost of rejuvenation is still high. Compared to rejuvenation, checkpointing is a light-level software fault-tolerance technique, the downtime cost of which is much less than that of rejuvenation. This paper presents a new idea to combine both these techniques to further reduce the total downtime cost of software systems. The Petri Net models of software systems and their corresponding numerical results are given in this paper.