When a cloud provider doesn’t properly schedule workloads, performance usually degrades. If the degradation is not caught in time, a cloud outage can occur. A couple of minutes of downtime is too long for many enterprise users to wait for the cloud service to resume. These cloud users would seek another cloud provider who could provide better service quality through proper workload allocation and scheduling. But finding the most optimal resource scheduling algorithm (RSA) to provide quality service is a challenging job.
The authors take this challenge by conducting a survey, according to the paper, “of 110 research papers out of [a] large collection of 1206 research papers published in 19 workshops, symposiums and conferences and 11 prominent journals.” One of the most interesting aspects of their survey is their description of RSAs based on a distributed resource scheduling strategy in a table format. For each algorithm, searching mechanism, application type, optimal, scheduling criteria, operational environment, merits and demerits, and technology are indicated.
As an example, the objective of the energy subtype of dynamic and adaptive RSA is to reduce power consumption. A central processing unit (CPU) intensive workload application running on Xen Hypervisor is used to reduce service-level agreement (SLA) violations, but the failure prediction is not measured accurately.
The authors keep the illustrations and tables informal to show the differences between resource scheduling algorithms in the cloud. For those interested in the issues and challenges of optimizing the algorithms, this paper is worth reading.