| Resource discovery is an important aspect of Computational Grids. Locating resources in a Grid environment is difficult because of the geographic dispersion and dynamic nature of its resources. Issues such as large numbers of users, heterogeneous resources and dynamic status of resources over time in a large distributed network make resource discovery more difficult than the case of traditional networks. In the case of Computational Grids, additional issues such as different operating systems, different administrative domains and lack of portability between platform dependent applications make resource discovery even more difficult. Further to all these difficult issues, knowledge of the current status of the resources adds an extra challenge to the problem of finding resources in the Grids. An ideal Computational Grid environment should contain a resource discovery infrastructure that includes heterogeneous resource monitoring capabilities. These capabilities will save time and the risk of selecting inappropriate resources.; In this thesis work, we propose a resource discovery infrastructure in the form of an automated status monitoring model. The model consists of two fundamental aspects, a portable data model and a set of executable monitoring components. Our approach adheres to principles of software design, is well structured and platform independent. The portable data model, which conveys the status of the resources, must be understandable by any application software, agent or scheduler on any platform. In turn, the monitors must be able to acquire necessary status information from various, diverse systems and maintain the data model. We developed appropriate interfaces that provide straightforward connectivity between our infrastructure and other Grid middleware components being developed elsewhere. |
