Abstract: High availability is the main features and the fundamental needs of mission-critical network system.Based on the analysis of system restart level and micro-reboot，a new judgement approach of best recovery strategy for mission-critical network system is presented，which is based on probability.The approach calculates recursively a variety of strategies needed to the restart recovery time according to failure probability of the system different levels，and then restarts the shortest time strategy as the best recovery strategy.Results of this case study show that the approach can reduce system restart recover time as much as possible and improve system availability.

Key words: reboot, mission-critical network system, high availability, best recovery strategy

摘要：

高可用性是任务关键网络系统的主要特征和根本需求。在分析系统重启层次及微重启技术的基础上，给出了一种基于概率的任务关键网络系统最佳恢复策略判定方法。该方法根据系统不同层次对象的失效概率，递归计算出各种策略所需要的重启恢复时间，然后将重启时间最短的策略作为最佳恢复策略。实例研究结果表明，该方法可尽可能地降低系统的重启恢复时间，提高系统的可用性。

关键词: 重启, 任务关键网络系统, 高可用性, 最佳恢复策略