摘要
针对高一致性应用的场景,基于云平台中间件提供的支持,从业务特性出发,设计了一种与数据层异步复制技术相结合的应用层容错机制,该机制保存状态数据的快照,在数据层异常时对未达到最终一致的状态数据进行访问控制,并允许其他状态数据继续进行业务操作。实验证明,该机制在保证数据一致的前提下,有效缩短了系统的故障切换时间,提高了系统整体的可用性,并对系统吞吐量不造成太大影响。
According to highly consistent scenarios, this paperproposes modular design of a fault tolerance mechanism in application layer from the perspective of business characteristicscombinedwith asynchronous replication technology in data layer based on the support provided by cloud middleware. This mechanism takes a snapshot of state data. When the data layer is down, the mechanism forbids access to the inconsistent data and allows business operation on the consistent data. Testing results show that the fault tolerance mechanism shortens the failover time effectively and improve the system＇s availability under the premise of guaranteeing data consistency. At the same time, it doesn＇t have an obvious impact on the system＇s throughput.
出处
《微型电脑应用》
2016年第2期63-68,共6页
Microcomputer Applications
关键词
云平台中间件容错机制一致性
可用性
Cloud Middleware
Fault Tolerance Mechanism
Consistency
Availability
