Minimising energy costs of data centers using high dense heterogeneous systems and intelligent resource management

Abstract

Recent trends in data centers include on one hand consolidation in large data centers and on the other hand increased number of small edge systems located closer to sources of data or energy. In both cases the key features needed to minimise costs are efficiency, adaptation to changing load and external conditions, and highly autonomous management. Additionally, especially in small edge data centers, efficiency and performance can be often optimized for specific applications. However, dynamic management and achieving efficiency gains for platforms containing a significant number of heterogeneous server nodes in a limited space require appropriate management methods. For example, suspending individual components and proper allocation of nodes becomes crucial to achieve high efficiency. In this paper we present a microserver system and demonstrate how much data center operator can gain in terms of overall costs by applying such systems with a special focus on intelligent power management techniques. In particular, we propose the novel power capping method that takes into account dynamic priorities, efficiency of specific types of heterogeneous server nodes, and impact of power capping on efficiency of specific applications. In this way, it both keeps the requested power usage levels as well as to reduce the overall energy consumption. Proposed architecture and management methods jointly allow to save around 30% of energy costs in a data center.