论文部分内容阅读
在大型计算机存储系统中,按模交叉存取工作方式,需要多个存储体同时工作,故存储系统的可靠性要求非常高,为了满足这一可靠性要求,通常在可靠性设计上,除了尽可能提高每个存储体的可靠性外,还采取在存储体上加海明码和遇体上发生两位错或其它系统错误时用备份体替代的办法来提高系统可靠性。本文指出上述提高系统可靠性的办法不是最佳的,只有采取本文给出的容错措施才是使容错能力达到最佳的存储系统。通过计算表明,它的一位错可靠性可以比原先提高6.6倍,几乎与两位错的提高倍数6.9倍相接近。本文说明了提高系统可靠性的重要性,并为大型计算机上研究容错问题提供了一个实例。
In a large computer storage system, according to the work mode of cross-access by mode, multiple banks are required to work at the same time. Therefore, the reliability requirements of the storage system are very high. In order to meet this reliability requirement, usually in the reliability design, May increase the reliability of each bank, but also take the bank on the Hamamei code and encounter two dislocations or other system errors with the backup body to replace the way to improve system reliability. This paper points out that the above methods to improve the reliability of the system are not the best. Only by adopting the fault-tolerant measures given in this paper is the storage system with the best fault tolerance. The calculation shows that its one-bit reliability can be increased by 6.6 times compared with the original, almost 6.9 times the improvement of two-bit error. This article explains the importance of improving system reliability and provides an example of how to deal with fault tolerance on a large computer.