Performance Modeling and Energy Conservation in Cloud Computing

Abstract

Nowadays, cloud computing is a hot and promising technology close to daily life. Energy consumption of the cloud has become a burning issue for researchers all over the world. To minimize the energy cost as well as ensuring network performance in the same time, an efficient and pervasive scheduling algorithm is needed for cloud computing. In this report, a dynamic right-sizing algorithm (Lazy Capacity Provisioning) is introduced and tested based on a data center model using real traced data. The online LCP solutions are evaluated based on an offline optimal solution. Further discussions are done related to its cost reduction compared with static right-sizing algorithm, its running time complexity, and its performance during different load conditions. In conclusion, this scheduling algorithm can successfully balance the trade-off between energy cost and network performance in the cloud and can be applied to practical production and life.