Quantum Virtual Machine: A Scalable Model to Optimize Energy Savings and Resource Management

• André Felipe Monteiro UFF / CEFET/RJ
• Orlando Loques UFF

Resumo

This work presents a model for managing Virtual Application Servers (VAS) on cluster-based web servers. In addition to providing energy savings, our model has linear scalability and defines a default processing virtual web server, known as Quantum Virtual Machine (QVM). A set of QVM performs a Logical Web Server (LWS), which operates in a flexible manner, changing its performance and power consumption depending on the workload of the applications. Concepts of agile VAS clone, co-allocation of VAS in the same core, and Dynamic Voltage and Frequency Scaling (DVFS) are used in the model, enabling rapid configuration actions and a fine-grained QoS control. The experiments evaluate the effectiveness of the proposed model by means of power consumption reduction and QoS violation as compared to the Linux CPU governors and state-of-the-art approaches based on optimization. The results show our model conserves up to 51.8% of the energy required by a cluster designed for peak workload scenario, with a negligible impact on the applications performance.

Palavras-chave: Cloning, Web servers, Power demand, Optimization, Performance evaluation, Resource management, power-aware computing, resource management, scalability, virtualization, web server clusters