A delay scheduling algorithm in short operating environment

A delay scheduling algorithm in short operating environment

Daiju Dong Xiao She Zhu Zhengdong WY PK.

A delay scheduling algorithm based on local resource prediction in cloud computing environment (locality resource forecast delay is proposed for the delay scheduling algorithm based on static time waiting threshold and not reasonable waiting in the yarn platform under short job scene. scheduling, LRFD). This algorithm takes into account the characteristics of short operation and dynamic change of resource availability to carry out task scheduling. The local resource information of the job is estimated based on the completion progress of the task on the node and the distribution of the job's unhandled data in the cluster, so as to determine whether the waiting is needed to improve the system performance and to realize the reasonable waiting for the local resources. The experimental results show that the performance and stability of the LRFD algorithm are better than those of the existing ones in the short operation scene, and the performance is increased by about 10%, and the maximum acceleration ratio can reach 3 times times.

A delay scheduling algorithm in short operating environment