Temperature-Aware Virtual Data Center Embedding to Avoid Hot Spots in Data Centers

Abstract

Virtual Data Center (VDC) embedding has drawn significant attention recently because of the growing need for efficient and flexible means of Data Center (DC) resource allocation. Existing studies on VDC embedding mainly focus on improving DCs' resource utilization. However, an important problem that has not been considered in VDC embedding solutions is the creation of hot spots by excessive heat dissipation and hot air generation from racks in DCs, which have significant adverse effects on energy consumption of the cooling system and IT equipment lifespan. To address this issue, we propose a temperature-aware VDC embedding scheme to avoid hot spots by minimizing the maximum temperature of hot air drawn to each rack. Meanwhile, we also aim to reduce the total power consumption of IT equipment in this scheme. A Mixed Integer Linear Programming (MILP) model and a heuristic algorithm are developed to implement the proposed VDC embedding scheme. Numerical results show that the proposed temperature-aware embedding scheme can significantly outperform a load-balanced embedding scheme in terms of maximum rack temperature, total power consumption of IT equipment, and VDC rejection ratio. Compared with a scheme that focuses only on maximizing IT resource utilization, which we refer to as IT-only energy-efficient scheme, our proposed scheme can significantly reduce maximum rack temperature and VDC rejection ratio while keeping the total power consumption of IT equipment close to that of an IT-only energy-efficient scheme.