A customized dynamic array for job scheduling queue of a large-scale server cluster

As computing demands increase, the management and scheduling of a volume of jobs is becoming a significant challenge. This paper analyzes various data structures and algorithms used to manage job queue. Considering a fair and adjustable job scheduling, job scheduling algorithms are generally based on dynamic priorities of jobs. This paper analyzes three conventional data structures, including binary tree, Vector, linked list, and corresponding algorithms used to implement priority queue. Based on the application scenarios, we propose a customized dynamic array to manage priority queue, and design corresponding algorithm, including job queue sorting, inserting, removing and etc. A performance measurement and comparison of these algorithms are performed in varying job scale. The measurement results show time cost of sorting in the dynamic array grows slower than Vector and binary tree. Moreover, the customized dynamic array causes more economical space cost than Vector. Considering performance and actual requirement, the customized dynamic array is more competent for implementing a job scheduling queue than others.