Distributed Meta-Scheduling for Grids 15th Mardi Gras Conference on Distributed Applications • Baton Rouge, Louisiana • 2008

Grid computing, a special form of distributed computing, stands for the effort undertaken mainly by computing centers to open up and combine their resources for an enhanced availability. With the increasing size of these compute infrastructures, there is a growing demand for an automatic balance of inter-infrastructure resource requests. Existing middleware such as UNICORE and Globus Tool Kit is ill-suited to this job since it requires the user to provide the location of suitable resources and only facilitates the migration process. Other projects like Gridway or LSF Multicluster suffer (at least currently) from missing interoperability. We describe a distributed meta-scheduling architecture that allows the automatic exchange of jobs between resource providers, aiming at improved resource utilization, automatic load-balancing, as well as reduced turn-around times. Additionally, the architecture is resilient to link and site failures due to its distributed nature. Finally, the system tries to achieve grid-wide improvements while still preserving the autonomy of resource providers. This is accomplished by making all decisions locally