基于分布式架构的星载并行计算机容错技术

星载计算机需要容错技术来满足在外太空运行的可靠性要求。目前的星载计算机多机系统通常设计为主从结构,集中于一个主节点上进行容错策略控制,这种结构存在着一点失效即瘫痪的隐患。为此,本文提出一种分布式架构下的星载并行容错计算机系统,将集中控制的容错部件分布化于各个节点之上,提高了系统的容错可靠性,在此架构上提出了计算节点、容错部件和I/O等容错策略,并给出了相应的模型及模拟测试结果,为进行类似项目的开发研究提供了有价值的指导和参考。

Fault-Tolerance Techniques for On-Board Parallel Computer System Based on Distributed Architecture

Fault-tolerant techniques can provide high reliability for on-board computers running in the outer space,the current multi-node on-board systems are designed as a master-slave structure,which focuses on the strategy of fault-tolerance in the master node and hereby contains a hidden danger.A parallel fault-tolerant computer system with a distributed framework is proposed in this paper.Based on the framework,the computing nodes and fault-tolerant units are designed and some novel fault-tolerant strategies are introduced.Our work can serve as an important guideline for the development of the related projects.