A dynamic data-aware scheduling for map reduce in cloud

Cloud services are employed for different purposes such as storage, delivery and processing of data. The workloads encountered are mainly heterogeneous resource demands. Heterogeneous distributed systems are often considered as a combination of public and private cloud systems, mobile based clusters and networks. Many schedulers and scheduling algorithm lacks reliability requirements for tasks and execution fails due to allocation of tasks to incongruous resources. Intelligent resource utilization is the key to deal with variable demand loads and therefore scheduling algorithms should be able to employ effective resource utilization techniques and also exploit additional resources if there is demand in resources. We propose an intelligent data-aware scheduler which enhances data-locality scheduling and also provides fairness in shared heterogeneous workloads. We conducted a brief evaluation of our scheduler and performance maintains better performance even if the task length increases.