New tool created to judge Grids
Computer grids, otherwise known as 'distributed computing' are used to work on processing data intensive problems. They work by permitting the distributed processing of data and objects across a network of connected systems. It has developed as an alternative to the supercomputer. The future aim is that multiple and geographically dispersed research groups will have the capability to pool their computing and data management resources. Therefore, results can be generated that would not be possible if each were working alone. Practically this tool has been implemented on projects that try to predict global climate change, deal with earthquake monitoring, or develop drugs to combat cancer and AIDS. The internet raises many questions regarding the performance characterisation of Grid infrastructures. Benchmarking metrics published on the Grid can provide a basis for users to assess the "quality of service" expected by a Virtual Organization providing computational services at a given cost. Scientists based at the University of Cyprus have put together GridBench. It is designed to be a user friendly system and enables various performance and reliability analyses to be made. It facilitates the easy definition of parameters on the grid and allows for the archiving and retrieval of results. It is also possible to use the results to create customised charts and graphs. New benchmarks have been added to the already established high performance computing (HPC) benchmarks that have already been tried and tested. Expected users are grid end users and grid infrastructure operators. They can use GridBench to assess the performance and functionality of their resources.
