Improving the Eco-Efficiency of High Performance Computing Clusters Using EECluster

Abstract

As data and supercomputing centres increase their performance to improve service quality and target more ambitious challenges every day, their carbon footprint also continues to grow, and has already reached the magnitude of the aviation industry. Also, high power consumptions are building up to a remarkable bottleneck for the expansion of these infrastructures in economic terms due to the unavailability of sufficient energy sources. A substantial part of the problem is caused by current energy consumptions of High Performance Computing (HPC) clusters. To alleviate this situation, we present in this work EECluster, a tool that integrates with multiple open-source Resource Management Systems to significantly reduce the carbon footprint of clusters by improving their energy efficiency. EECluster implements a dynamic power management mechanism based on Computational Intelligence techniques by learning a set of rules through multi-criteria evolutionary algorithms. This approach enables cluster operators to find the optimal balance between a reduction in the cluster energy consumptions, service quality, and number of reconfigurations. Experimental studies using both synthetic and actual workloads from a real world cluster support the adoption of this tool to reduce the carbon footprint of HPC clusters.

On the Integrity of Performance Comparison for Evolutionary Multi-objective Optimisation Algorithms

“…The authors wish to determine the effectiveness of using a multi-objective evolutionary algorithm approach to solve the problem of how best to manage a robot. Similar experimental scenarios which use implementations from the MOEA framework can be found in [26] and [3].…”

On Algorithmic Descriptions and Software Implementations for Multi-objective Optimisation: A Comparative Study

“…It supports genetic algorithms, differential evolution, genetic programming, and more. A number of state-of-theart EMO algorithms, test problems, and quality indicators are provided out-of-the-box and have been used in multiple experiment studies detailed in the EMO literature [5,15,52].…”

Eco-Efficient Resource Management in HPC Clusters through Computer Intelligence Techniques

“…Five different solutions were tested in the experiments to implement the node allocation algorithm (second stage of the decision-making mechanism), while sharing the same implementation for the first stage in the form of the reactive strategy described in Section 4.1. Note that it is not the point of these experiments to assess the results of the slot allocation, as this has already been tested in previous works (see [45,46,49]). In particular, four node-ordering heuristics proposed in Reference [19] were tested along with our proposal:…”