Energy-Efficient Data Center Network Infrastructure With Network Switch Refresh Model

Abstract

Data centers and their underline transmissions are required to guarantee critical services in a 24/7/365 framework. The extensive energy consumption of these data centers and their transmission networks impose a threat to globally available scarce energy resources. Furthermore, addressing performance and energy requirements trade-offs is also challenging. This article has been specifically focused to assess the inside-energy-view of typical data center networks and assess how network infrastructure replacements or ''refresh'' can lead to a better energy-efficient data center network (DCN) design without compromising performance or service level requirements. Server refresh techniques are found in the literature. However, considering network infrastructure refresh to attain energy efficiency is the first of its kind. A model has been proposed in this article that works on algorithms to recommend network switch replacements. The algorithms perform parametric analysis for replacement. The analysis considers parameters impacting the performance and energy consumption of the switches. In addition to technical parameters, the proposed model has also evaluated the cost impacts of the replacement. The replacement is only recommended if it is energy efficient and performance effective and validated when replacement is also cost-effective. The proposed model has been evaluated through two replacement options, one with the same manufacturer and the other with a different manufacturer. Replacement with two different options has been proposed to evaluate the impact of various manufacturers in designing a better energy-efficient data center meeting all service level requirements and performance guarantees. The results obtained from the proposed reflect the attainment of the desired objectives.