Cluster resource scheduling in cloud computing: literature review and research challenges

Scheduling plays a pivotal role in cloud computing systems. Designing an efficient scheduler is a challenging task. The challenge comes from several aspects, including the multi-dimensionality of resource demands, heterogeneity of jobs, diversity of computing resources, and fairness between multiple tenants sharing the cluster. This survey provides a multi-perspective overview of the cluster scheduling problem. We present a multi-dimensional classification of existing cluster management solutions based on their scheduling architectures, objectives, and methods. We also survey the recent research works which have employed machine learning solutions in cloud computing resource management. Existing cluster scheduling systems face many challenges, such as achieving a tradeoff between multiple conflicting objectives, finding the balance between jobs' requirements, scaling to the new operational demands, and choosing the appropriate scheduling architecture. Using machine learning in cluster scheduling is a promising direction to go to develop the future generation of intelligent resource schedulers.