Introduction
The real problem underlying the Grid concept for advanced e-Science applications is coordinated resource sharing and problem solving in dynamic, multi-institutional virtual organisations (VOs). Previous attempts and existing approaches for such large-scale distributed applications are often tightly coupled, static, inflexible, and associated with specific protocols. The overall goal of the e-Demand project is to address this problem by developing a demand-led and service-centric architecture for building complex but dependable and secure Grid applications based on the notion of ultra-late binding, dynamically bound service components, combined with atomic actions as a powerful control abstraction. This new architecture is protocol-independent, and able to wrap and package other applications and technology including existing Grid protocols and services.
Three new generic services will be developed to enhance the architecture, which will provide support for 1) coordinated multi-party interactions, 2) attack detection and tolerance and 3) stereoscopic visualisation. The e-Demand solution represents the significant trend in the application space - future Grid applications will be based on compositions of services discovered and provided dynamically at run time (i.e. just-in-time integration of services).
The University of Leeds Distributed Systems and Services (DSS) group has had for many years an established track record and an internationally leading reputation in service-based software architectures, dependable distributed computing and software visualisation. The Leeds and Durham Grid consortium, including experts from both academia and industry (e.g. Sun, Sharp, and Sparkle Computer Technology) will act as a research team to lay the foundations for the project and develop solutions to the Grid problem. The project will also benefit from direct consultation from the north-east regional e-Science centre.
A Conceptual Service-Based Architecture , associated with new services, is the core of the proposed research:
The research program consists of three major stages: i) develop an architectural model and implement an instance of the new architecture, ii) within the architectural framework, develop and implement three high-level Grid services, and iii) integrate them into a workable system, and test and evaluate the system with fault injection and two case studies. This project will fund two RAs (two years each) and a three-years PhD studentship. It will also fund a networked Sun system (including servers and workstations), Sharp Stereo Displays, and the Access Grid system (video conferencing).
|
