Preliminary Evaluation of Performance and Financial Cost of HPC Applications on AWS Burstable Instance Clusters

• Artur Soda UFSC
• Vanderlei Filho UFSC
• Márcio Castro UFSC

DOI: https://doi.org/10.5753/eradrs.2025.6397

Abstract

Amazon Web Services (AWS) Burstable Instances offer variable performance based on CPU credit consumption, which can present challenges in evaluating the cost-benefit for HPC workloads. This paper analyzes the performance and financial cost of HPC applications on clusters with both Non-Burstable and Burstable instances. Our results show that the choice of the ideal instance type depends on the workload profile, with Non-Burstable instances being more efficient for intensive workloads and Burstable Standard being suitable for light workloads.

Keywords: High-Performance Computing Applications in Sciences and Engineering, Computing in Computer Clusters, High-Performance Computing in Grid and Cloud

References

Bailey, D. H., Barszcz, E., Barton, J. T., Browning, D. S., Carter, R. L., Dagum, L., Fatoohi, R. A., Frederickson, P. O., Lasinski, T. A., Schreiber, R. S., Simon, H. D., Venkatakrishnan, V., and Weeratunga, S. K. (1991). The nas parallel benchmarks. The International Journal of High Performance Computing Applications, 5(3):63–73.

Ferrari, G., Filho, V., and Castro, M. (2024). Comparing burstable and on-demand aws ec2 instances using nas parallel benchmarks. In Anais da XXIV Escola Regional de Alto Desempenho da Regiao Sul, pages 61–64, Porto Alegre, RS, Brasil. SBC.

Leitner, P. and Scheuner, J. (2015). Bursting with possibilities – an empirical study of credit-based bursting cloud instance types. In 2015 IEEE/ACM 8th International Conference on Utility and Cloud Computing (UCC), pages 227–236.

Munhoz, V. and Castro, M. (2023). Enabling the Execution of HPC Applications on Public Clouds with HPC@Cloud Toolkit. Concurrency and Computation: Practice and Experience, pages 1–19.