Large-scale Distributed Processing System Using Loop Networks
スポンサーリンク
概要
- 論文の詳細を見る
Distributed processing is attracting attention as a method for configuring information processing systems which can provide high reliability while facilitating system expansion. The DIPS system introduced in this paper is designed on a fully integrated-hardware to application software-concept and employs general-purpose computers to perform large-scale distributed processing functions. Featuring a high-performance optical loop, this network provides the system expandability and interprocessor communication reliability required by the information processing services of the 1990s. Problems related to large-scale distributed processing system configuration and methods introduced for solving them are described in this paper along with an evaluation of these methods based on their application results. The following methods in particular are described:1) An innovative processor distribution configuration which distributes the communication control function all the way to the upper levels and improves system reliability and cost effectiveness,2) An interprocessor communication method using a high-performance optical loop for providing system expandability and high reliability,3) An automated operation scheme for large-scale distributed processing systems which uses system control processors,4) A method which exploits the inherent characteristics of distributed processing-the ability, for example, to mask host processor failures from the terminals-to improve reliability, and 5) A method for expanding the system without shutting down online services.
- 一般社団法人情報処理学会の論文
- 1992-03-15
著者
-
Nakatani Hajime
Ntt Network Information Systems Laboratories
-
Kinouchi Y
Tokyo Univ. Information Sciences
-
TSURUHO SEISHIRO
NTT Software Laboratories
-
KINOUCHI YASUO
NTT Communication Switching Laboratories
-
HOSHIKO TAKAYUKI
NTT Network Systems Development Center
-
MIYAKAWA JUNJI
NTT Affiliated Companies Department