基于仲裁的高可用服务器故障检测机制的研究

故障检测是设计高可用系统的一项关键技术，本文研究了双模高可用容错服务器的故障检测机制。首先，通过马尔科夫模型对服务器系统进行了分析，指出设计高覆盖率和高成功率的故障检测机制对系统可用性的重要影响。针对传统使用的故障检测机制成功率较低的问题，提出了一种仲裁检测机制，仲裁系统采用容错方案设计，具有较高的可靠性，在双机无法对故障做出正确判断时，可作为可信的第三方对故障做出准确的定位，有效地提高了系统的检测成功率。在此基础上，一种基于仲裁的多层心跳检测机制被应用到了实际系统的设计中，并通过试验证明了其可满足可用性设计要求。

The Research on Fault Detection Based on Arbitration in Highly Available Server

Fault detection is one of the key technologies of designing highly available systems. This paper studied the fault detection mechanism of the dual fault tolerance server. First, we used Markov model to analyze the system of server and pointed out the important influences of fault detection mechanism with high coverage and success rate to the availability of system. Aiming at the lower successful fault detection in traditional system, a detection mechanism based on arbitration was proposed. The Arbitrator can reach very high reliability due to adopting design of fault tolerance, when dual machines cannot make right decision to the fault, the arbitrator, as a dependable third party, can locate the fault exactly, which improves effectively the success rate of fault detention. Basing on this, a kind of hierarchical fault detection mechanism based on arbitration was applied to the design of practical system, and proved by experiments that it can satisfy requirement of availability.