Hierarchical model-based associate tasks scheduling with the deadline constraints in the cloud

Abstract

Cloud computing can provide the dynamic and elastic virtualized resources for the users and is based on distributed computing, parallel computing and grid computing. Task scheduling is an important part of cloud computing. The schedule constraint is based on the QoS constraints, such as task executed time, cost, resource utilization, etc. We proposed one task hierarchical model for the associated task scheduling considering the real application requirements in cloud computing. Considering the parallel structure of sub-DAG, we proposed the hierarchical task graph to decompose the associated tasks, which can improve the tasks execution concurrency and reduce the execution cost. In order to execute all of the associated tasks in the specific delay-bound, we proposed the concept of tasks processing capacity and the corresponding calculation method, and further established the mapping between the task processing capacity and execution time. Concerning the delay of the associated tasks scheduling in cloud computing, the associated task scheduling algorithms based on delay-bound constraint based on the task hierarchical model was proposed. The scheduling algorithm is called associated tasks scheduling based on serial/parallel structure (SAH-DB). Extensive experimental results demonstrated that the proposed SAH-DB algorithms can achieve better performance than CPM and TS-Sim algorithm in the terms of the total execution cost and resource utilization.