Abstract: This paper presents a data allocation strategy used in High Performance Data Processing Middleware(HPDPM) system to improve system parallelism, reduce communication costs, make the inquiry load price to be smallest, and enhance system response ability. Based on the fundamental thought and working principle of this strategy, a framework and the primary modules of the middleware system are described. This middleware system is applied to a national large engineering project.

Key words: parallel computing, database system, middleware, storage allocation

摘要： 提出一种用于高性能数据处理中间件系统的数据分配策略，以提高系统并行性、减少通信开销、使总体查询负载代价最小并增强系统响应能力。基于该策略的基本思想和工作原理，给出中间件系统的体系结构和主要功能模块，通过实验证明其有效性。该系统已在一项国家大型工程中得到应用。

关键词: 并行计算, 数据库系统, 中间件, 存储分配

CLC Number: 

• TP311.13