AutoSaver: Cost-driven Resource Allocation for Multi-tier Web Application in Virtualization Environment
Shuguo Yang, Haixia Li, Xiong Pengcheng · 2012
This paper presents Auto Saver to balance the performance and resource utilization for a web application in virtualization environment. It consists of a feedback utilization controller, an ARMA-based model estimator and an optimizer. Experiment results show that the controller can maintain the utilization, the estimator can track the relationship between metrics and control knobs, and the optimizer can make trade-offs between performance and resource of the web application.