Skip to Main Content
The resources needed to execute workflows in a Grid environment are commonly highly distributed, heterogeneous, and managed by different organizations. One of the main challenges in the development of Grid infrastructure services is the effective management of those resources in such a way that much of the heterogeneity is hidden from the end-user. This requires the ability to orchestrate the use of various resources of different types. Grid production infrastructures such as EGEE, DEISA and TeraGrid allow sharing of heterogeneous resources in order for supporting large-scale experiments. The interoperability of these infrastructures and the middleware stacks enabling applications migration and/or aggregating the combined resources of these infrastructures, are of particular importance to facilitate a grid with global reach. The relevance of current and emerging standards for such interoperability is the goal of many researches in the recent years. Starting from a first prototype of grid workflow, we have improved the current version introducing several features such as fault tolerance, security and a different management of the jobs. So our grid meta scheduler, named GMS (Grid Meta Scheduler), has been redesigned, able to be interoperable among grid middleware (gLite, Unicore and Globus) when executing workflows. It allows the composition of batch, parameter sweep and MPI based jobs.