Title: Hybrelastic: a hybrid elasticity strategy with dynamic thresholds for microservice-based cloud applications

Authors: José Augusto Accorsi; Rodrigo da Rosa Righi; Vinicius Facco Rodrigues; Cristiano André da Costa; Dhananjay Singh

Addresses: Applied Computing Graduate Program, Universidade do Vale do Rio dos Sinos, São Leopoldo, RS, Brazil ' Applied Computing Graduate Program, Universidade do Vale do Rio dos Sinos, São Leopoldo, RS, Brazil ' Applied Computing Graduate Program, Universidade do Vale do Rio dos Sinos, São Leopoldo, RS, Brazil ' Applied Computing Graduate Program, Universidade do Vale do Rio dos Sinos, São Leopoldo, RS, Brazil ' Department of Electronics Engineering, Hankuk University of Foreign Studies, Yongin 17035, South Korea

Abstract: Microservices-based architectures aim to divide the application's functionality into small services so that each one of them can be scaled, managed, implemented, and updated individually. Currently, more and more microservices are used in application modelling, making them compatible with resource elasticity. In the literature, solutions employ elasticity to improve application performance; however, most of them are based on CPU utilisation metrics and only on reactive elasticity. In this context, this article proposes the hybrelastic model, which combines reactive and proactive elasticity with dynamically calculated thresholds for CPU and network metrics. The article presents three contributions in the context of microservices: 1) combination of two elasticity policies; 2) use of more than one elasticity evaluation metric; 3) use of dynamic thresholds to trigger elasticity. Experiments with hybrelastic demonstrate 10.31% higher performance and 20.28% lower cost compared to other executions without hybrelastic.

Keywords: elasticity; reactive elasticity; proactive elasticity; scalability; dynamic thresholds; microservices.

DOI: 10.1504/IJCC.2024.137402

International Journal of Cloud Computing, 2024 Vol.13 No.2, pp.99 - 123

Received: 07 Dec 2021
Accepted: 03 May 2022
Published online: 18 Mar 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article