A dynamic delay optimization scheduling model

The response time starts to suffer due to the limitation of the Hadoop’s FIFO scheduler, and this is unacceptable to execute the large scale instance intensive tasks. To enhance the system resource utilization, we propose a new scheduling solution. To reduce the cost, we use a delay scheduling algorithm to determine the scheduling opportunity. Delay scheduling can ensure that the current service can make full use of the resources, improve resource utilization, and reduce the probability of failure scheduling. The initial experiments demonstrate that the large scale instances intensive workflow tasks will benefit from the Min-Cost delay scheduling algorithm that is proposed in this paper. © Springer International Publishing Switzerland 2014.