The goal of this NSF REU project is to understand and improve resource management and job scheduling on large-scale clusters. Major tasks include literature survey, exploring Omega-based cluster scheduler simulator with workload traces from Google production systems, and developing new scheduling policies.