Shiow-Luan Wang, Yung-Tsung Hou
The aim of this study is to present a distributed batch job scheduling problem that is practical in big data analysis. When a computer cluster receives computing jobs from different sources, different jobs can be combined into a time slot for batch processing. The cluster has multiple servers for batch processing, and a computing job can be split into different time slots on a server. The objective of the scheduling is to minimize the total completion time of all jobs. We show that this problem is NP hard, and this paper also proposes a (2−1m) approximation algorithm for this batch scheduling problem, where m is the number of servers, and follows up the reliability evaluation to make the system more stable and reliable.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados