On revisiting energy and performance in microservices applications: A cloud elasticity-driven approach
Fontana de Nardin I, da Rosa Righi R, Lima Lopes TR, André da Costa C, Yeom HY, Köstler H (2021)
Publication Type: Journal article
Publication year: 2021
Journal
Book Volume: 108
Article Number: 102858
DOI: 10.1016/j.parco.2021.102858
Abstract
Monolithic applications are a subject that includes several knowledge areas. Sometimes it can be a challenge to optimize CPU or IO requirements because it is not trivial to recognize the problem itself and improve it. There are many approaches to resolve this situation, where a trending one is the microservices. As a variant of the service-oriented architecture, microservices is a technique that arranges an application as a collection of loosely coupled services. This decomposition enables better software management in cloud-based environments since we can replicate each part individually using cloud elasticity to avoid execution bottlenecks. Also, since elasticity mitigates resource overprovisioning, it favors better energy consumption: the cloud owner can redistribute finite available resources among different tenants, and users can pay less to use the infrastructure. However, elasticity tuning is not trivial and depends on several factors, such as user experience, application architecture, and parameter modeling. Today, we observe a lack of initiatives in the literature that address both performance and energy perspectives to support the execution of microservices applications in the cloud. Concerning this context, this article introduces Elergy as a lightweight proactive elasticity model that provides resource reorganization for a cloud-based microservices application. Its differential approach appears in improving energy consumption by periodically handling the most appropriate amount of resources to execute an application while maintaining or yet improving the performance of CPU-bound applications. Elergy performs these functions proactively, in such a way of preventing future problems related to either resource under- or overprovisioning. The results showed energy consumption reduction and a competitive cost (application time x consumed resources) when comparing Elergy with a non-elastic scenario. Elergy obtained savings from 1.93% to 27.92% for energy consumption.
Authors with CRIS profile
Involved external institutions
Brazil (BR)How to cite
APA:
Fontana de Nardin, I., da Rosa Righi, R., Lima Lopes, T.R., André da Costa, C., Yeom, H.Y., & Köstler, H. (2021). On revisiting energy and performance in microservices applications: A cloud elasticity-driven approach. Parallel Computing, 108. https://dx.doi.org/10.1016/j.parco.2021.102858
MLA:
Fontana de Nardin, Igor, et al. "On revisiting energy and performance in microservices applications: A cloud elasticity-driven approach." Parallel Computing 108 (2021).
BibTeX: Download