Design and implementation of an efficient VM scheduling framework for interactive streaming service

Abstract

Cloud computing has become widely used to provide many services such as analyzing and streaming data to increase scalability and minimize up-front IT infrastructure costs. However, to make the best use of cloud infrastructures in terms of performance and cost, efficient virtual machine (VM) management is required. In this article, we propose a VM scheduling framework for automatic and cost-effective management of VMs for streaming services. The framework controls and manages the life-cycle and status of multiple VMs in the cloud platform automatically. In addition, we reduce VM response time by applying a VM scheduling policy (e.g., LRU algorithm) based on content usage. We implement the VM scheduling framework based on the Google Cloud platform (GCP). The experimental results show that the streaming services based on the proposed framework can provide lower costs with fewer performance overheads than the streaming services without the framework.