Resource Directory Information Service In Computational Grid Environment

As a latest emerging network infrastructure, Grid is the key technology of the new generation of the Internet. Along with the huge improvement and rapid progress of computer performance and network communicating technology, the development of application demands toward large-scale high-performance, diversity and multi-function needs to connect the various high-performance computing resources, storage, data and other specific resources, which are geographically distributed and heterogeneous, via high speed network to establish collaborate high-performance computing and solve huge application problems cooperatively, hence the name wide-area high-performance meta computing technology, Grid Computing Technology as well. One of the important issues of Grid Computing is a Resource-related one. The normal resources in Grid are: processing capability, storage resources, directories, network resources, distributed file system, distributed computing pool, computing clusters etc. In Grid Computing environment, resources are diverse among different administrative domain at various locations, which belong to different organizations and also different in deploy and security strategies. Most of the resources are inter-connected via the Internet. The resource operation under this scenario includes the organization, accession, localization, discovery, scheduling, allocation, authentication, and activities of process-creation as well as all the other activities about its preparation for resources. The approach to effectively discovering and inquiring resource information through different administrative domains stands an important sense to the scheduling and execution of Grid applications. Resource information service plays a decisive role in Grid computing environment, which is a fundamental basis. The functions and characteristics of the system should contain: having access to static and dynamic resource information of components, being able to arrange and adjust the configurable service according to heterogeneous and dynamical environment, unified effective implementation interface of accessing information, extensible accessing to dynamic data, having access to various resource and mode of administration based on distribution. All the services above have greatly influenced and thoroughly depend on what the information providers report to us, such as resource availability, capacity, usage, etc. Meta-computing directory service (MDS) is an extremely important component in Globus project as part of information infrastructure, which mainly contains resource discovery, resource description and resource monitoring & renewing. Using LDAP as the unified accessing interface of Grid information, MDS is now able to provide users with such information as the existing resource information, status information, and optimized information of Grid applications. MDS supports the creation and collection of Virtual Organizations and represents enormous resources from a standard point of view. We can establish the distributed resource information services hierarchically under the structure of GIIS/GRIS, which is an LDAP hierarchy one, facilitate the MDS with excellent distributed extensibility. It allows Effective multi-layer service structure such as hierarchical information service, coordinated information service and hybrid one. In Grid Computing environment, the development of efficient Grid resource information service that provides good extensibility, low-cost and abundant accurate information brings practical sense to effective scheduling, delivery and execution of applications in such environment. We have compared and analyzed the service performance of MDS through in-depth research and practical work on it, focusing on which we have proposed several analyses of the design of optimization along with concrete implementation of such strategies, including performance analysis of directory service, quality of MDS service, and analyses of Cache strategy and efficient issue of inquiring security. What's more, emphasizing on classified resource, we implement a multi resource information server at an index level, which has greatly improved the resource inquire and collecting performance, shortened the feedback response time of resource information at a basic level. Another issue to which we put much more attention is the realization of the integration of MDS and NWS based on analysis and practice of the MDS mechanism, by encapsulating NWS service into a component and backend of Grid information service, registered to the MDS frame. The integration stands upon LDAP hierarchical structure, announces NWS measurement data via a form of resource information service, which is a major way of integrating to the directory information service infrastructure. Considering MDS'not providing with network bandwidth and latency and the unreasonable architect of dynamic resource information, we enrich the network bandwidth and CPU parameters measurement into the structure, make a great improvement of the lacking information of dynamic ones as well as to the MDS service performance. Meanwhile we establish the function of prediction of resource information, which means predicting the resource promising performance in certain period of time, and it will do a perfect preparing job at the prophase of application scheduling and future system utilizations.