基于校园网络的元计算实验系统WADE的设计与实现

元计算系统是可以作为虚拟的整体而使用的地理上分散的异构计算资源，这些资源包括计算机、数据库和昂贵仪器等．元计算系统在硬件和软件方面均有异构性，适合具有不同内在并行性的复杂应用的执行．现存的绝大多数并行系统都是同构的，不具有这一优势，因此，研究异构环境下的元计算系统就很有现实意义．WADE是基于校园网络开发的元计算实验系统，使用MD支持异构数据格式转换，使用面向对象技术实现单一映像系统，使用优先约束的任务调度算法来实现应用程序的调度和运行，并提供与流行的并行编程软件如PVM等的接口。

DESIGN AND IMPLEMENTATION OF A CAMPUS-WIDEMETACOMPUTING SYSTEM(WADE)

A metacomputing system is a set of geographically distributed resources, including hosts, databases, expensive instruments, etc ., interconnected by high speed networks. These resources are heterogeneous in many aspects and may best support the applications with different inherent parallelism. Most existing parallel systems are homogeneous or identical, and have no such advantage. Therefore, it is practically meaningful to develop heterogeneous metacomputing systems. WADE is a prototype of a campus wide metacomputing system. WADE is characterized by a communication model for transparent conversion between heterogeneous data formats using meta data information, a single image space with uniform programming interfaces implemented with object oriented techniques, task scheduling algorithm with precedence constraints for application scheduling in the heterogeneous environment, and the interfaces to popular parallel programming systems like PVM.