Spatial applications will gain high complexity as the volume of spatial data increases rapidly. A suitable data processing and computing infrastructure for spatial applications needs to be established. Over the past decade, grid has become a powerful computing environment for data intensive and computing intensive applications. Integrating grid computing with spatial data processing technology, the authors designed a spatial data processing grid (called SDPG) to address the related problems. Requirements of spatial applica are examined and the architecture of SDPG is described in in the paper. Key technologies for implementing SDPG are discussed with emphasis.

This paper presents a novel approach to achieve high scalability and availability in a server environment based on operating system level Virtual Machine (VM). This approach mainly uses VM technology and Maximum Amount and Most Available Service Capaciry Finding (MA2SCF) algorithm designed by us to achieve Service Capacity Sharing (SCS) for overloading relaxation and fault management in a server environment. And we have implemented a prototype system including two software components VMMaster and VMAgent to interact with the server environment and VMS, and to automatically manage the SCS process. The current implementation has been tested in Linux Virtztal Server environment (LVS) and the approach itsel. can be applied to a more general server environment, and even a distributed or grid environment }vhich are our ongoing work. The experiment we present in this paper proves the. feasibility and effectiveness of oztr approach.

内存密集型应用对运行环境的物理内存要求严格，在物理内存不足时将会引发大量磁盘IO，降低系统性能。传统的网络内存致力于在集群内部通过共享空闲节点的物理内存解决该问题，但受集群负载和内部网络影响较大。通过结合网络内存和服务计算、网格计算等技术，提出一种基于内存服务的内存共享网格系统－内存网格，并分析和讨论了实现内存服务的关键技术和算法，内存网格弥补了网络内存的不足，扩展了网格计算的应用范围，通过基于真实应用运行状态的模拟，证明了内存网格与网络内存相比具有性能的提高。

作为面向互联网资源共享的虚拟计算环境的实例，iVCE（Internet based virtual computing environment） for Memory 致力于解决广域分布的内存资源的共享与综合利用问题。由于内存资源的特殊性，传统的资源管理方法很难适用。以iVCE for Memory作为背景，提出一种基于聚类的虚拟计算环境资源聚合方法，有效降低了资源聚合的问题规模；借鉴物理学中的力场和势能理论，建立了实现资源聚合的基本模型和力场-势能模型以及相应的分布式算 法；通过基于真实网络拓扑的模拟，对两种模型和算法分别进行了评估和验证。