网格环境下RPC的研究

Study on Grid-based RPC

网格技术是在巨型计算机与互联网技术的基础上推出的一项新变革,它将分布在各地的计算机资源包括、CPU存储器、数据库等,通过高速的互联网组成充分共享的资源集成。笼统地讲,的作用是将各种计算机联结起来,Internet网站是将各信息资源联结起来,而网格则是将各种计算机资源联结起来。网格应用环境具有并行性、分布性、异构性及动态性等特点,而且它需要在动态的环境下动态灵活地组合资源和服务。在广域范围内,分布式网格环境下容错机制是非常必要和基本的。网格的客户端和服务器端必须表现出独立的失败特性,因此,要达到这个目标,网格就需RPCRPC要使用事件和消息服务处理发生的错误。文章出于网格环境下的需要远程过程调用机制的目的,探讨与传统的的不RPC同,给出网格环境的的RPC机制和特点。首先描述基本的模型并列出可能中发生的错误。然后定义在网格RPC4中4个基本的操作,最后研究网格RPC的需求。

The innovation of grid technology is introduced based on the technology of giant computer and Internet. Grid forms computer resources distributed everywhere into integration of sharing sufficiently resources including CPU, memory, database etc. through high-speed Internet. Ambiguously saying, the function of Internet connects all kinds of computers; that of Web- site connects all kinds of information resources; while that of grid connects all kinds computer resources. A grid application environment is parallel, distributed, heterogeneous and dynamic etc. Furthermore, it will need to dynamically and flexibly compose resources and services across dynamic environment. In a wide-area, the need for fault tolerance is essential in the distributed environment. Grid RPC clients and servers must exhibit independent failures. Hence, to accomplish this, grid RPC will have to use an event and messaging service to manage the occurrence of errors. This document is motivated by the need for a remote procedure call mechanism for grid environments, and the difference compared to traditional RPC is presented, as well as the mechanism and characters in grid environments. The document begins by presenting a basic RPC model and enumerating possiblefaults in grid RPC. Four basic RPC operations are defined. Finally, the requirements for grid RPC are dicussed.