Abstract: Existing research achievements mostly focus on service pool model and service selection algorithms.Excessive redundancy and frequency adjusting service pool pay more costs.For minimize the costs of redundancy service，a dynamic redundancy service strategy is presented in this paper.It analyzes the availability fluctuation of service in service composition.Mutant point of service availability is obtained by grey model.It dynamically adjusts the redundancy service by multistage dynamic planning.The experimental results show that proposed dynamic redundancy service strategy can assure availability and decrease costs.

Key words: redundancy solutions, availability, service, multistage dynamic planning

摘要： 现有关于服务冗余策略的研究工作侧重于研究冗余服务池模型和服务选择机制，过度冗余或服务池频繁调整的开销较大。为了降低冗余服务的开销，提出了服务的多阶段动态冗余策略，在分析服务可用性波动特性的基础上，采用灰模型获取可用性突变点，并根据突变点对可用性进行阶段划分，根据动态多阶段规划确定服务冗余度。实验结果表明提出的多阶段动态冗余策略能够保障可用性并有效降低开销。

关键词: 冗余策略, 可用性, 服务, 多阶段动态规划