Achieving Fairness-Aware Two-Level Scheduling for Heterogeneous Distributed Systems

In a heterogeneous distributed system composed of various types computing platforms such as supercomputers, grids, and clouds, two-level scheduling approach can be used to effectively distribute resources the users in first-level, map tasks nodes for each platform second-level executing many-task applications. When resources, service providers should consider fairness among multiple well efficiency. However, cannot achieved by simply distributing an equal amount from every user. this paper, we investigate how address issue system. We present three first-level resource allocation policies provider affinity first policy, application based round-robin two task mapping most affected policy co-runner policy. Using trace-based simulations, evaluate performance combinations second level policies. Our extensive simulation results demonstrate that plays crucial role achieve relatively good fairness.