摘要
从判断一个节点好坏的全过程来看,利用传统心跳故障检测器的算法还是离不开超时技术。借助于心跳器的“计数”思路,提出了一个扩展心跳故障检测器模型,允许对计数器施行递增和递减操作,心跳序列随时间上下波动,从而输出更多的用于判断节点好坏的线索,使得基于扩展心跳故障检测器的算法彻底地摆脱了超时的阴影。
Judging from the process of distinguishing crashed nodes from correct ones, we find that the algorithm with the existing heartbeat failure detector can do nothing without time-out technology. By adopting the ways and means of "counting" of existing heartbeat failure detector, an extended heartbeat failure detector model is introduced, which allows addition and subtraction on the counters. The heartbeat curve stretches up and down according to time, consequently. As a result, more hints used to distinguishing crashed nodes from correct ones are outputted, and the algorithm with extended heartbeat failure detector shakes offtime-out technology completely.
出处
《计算机工程与设计》
CSCD
北大核心
2006年第21期4147-4149,共3页
Computer Engineering and Design
关键词
分布式系统
扩展心跳故障检测器
一致问题
异步系统
超时技术
算法
distributed system
extended heartbeat failure detector
consensus
asynchronous system
overtime technology
arithmetic