QoS-aware online mechanism for dynamic VM provisioning in cloud market using Q-learning

摘要

Cloud provider (CP) leases various resources such as CPUs, memory and storage in the form of virtual machine (VM) instances to clients over internet. This paper tackles the issue of quality of service (QoS) provisioning in cloud environment. We examine using Q-learning for provisioning VMs in the cloud market. The extracted decision function should decide when rejecting new request for VMs that violate QoS guarantee. This problem requires the reward for CP be maximised while simultaneously meeting a quality of service (QoS) constraints. These complex contradicting objectives are embedded in our Q-learning model that is developed and implemented as shown in this paper. Numerical analysis shows the ability of our solution to earn significantly higher revenue than alternatives.