View in EDS

On the Stability of the Kubernetes Horizontal Autoscaler Control Loop

Saved in:
Bibliographic Details
Title: On the Stability of the Kubernetes Horizontal Autoscaler Control Loop
Authors: Berta Serracanta, Andor Lukács, Alberto Rodriguez-Natal, Albert Cabellos, Gábor Rétvári
Source: UPCommons. Portal del coneixement obert de la UPC
Universitat Politècnica de Catalunya (UPC)
IEEE Access, Vol 13, Pp 7160-7166 (2025)
Publisher Information: Institute of Electrical and Electronics Engineers (IEEE), 2025.
Publication Year: 2025
Subject Terms: numerical simulations, Cloud autoscaling, System stability, microservices architecture, Àrees temàtiques de la UPC::Informàtica::Arquitectura de computadors, Control theory, Microservices architecture, Numerical simulations, Electrical engineering. Electronics. Nuclear engineering, Kubernetes, control theory, TK1-9971, Horizontal Pod Autoscaler
Description: Kubernetes is a widely used platform for deploying and managing containerized applications due to its efficient elastic capabilities. The Horizontal Pod Autoscaler (HPA) in Kubernetes independently adjusts the number of pods for each service, yet these services often operate in an interconnected manner. This study aims to understand the effects of autoscaling events on a graph of interconnected services. To achieve this, we apply control theory to model the HPA’s behavior. We analyze the stability of this model, perform numerical simulations, and deploy a real testbed to evaluate the performance. Our findings demonstrate that the control theory-based model accurately predicts the HPA’s behavior, ensuring system stability with CPU utilization meeting desired thresholds and no traffic loss after a transitional period. The model provides insights into optimizing resource scheduling and improving application performance in Kubernetes environments. Additionally, we extend our model to the whole service graph to understand how individual scaling decisions influence the complex graphs of cloud applications.
This work was supported in part by the Spanish I+D+i Project TowaRds fully AI-empowered NetwoRks, subproject A (TRAINER-A), funded by Ministerio de Ciencia e Innovación (MCIN)/Agencia Estatal de Investigación (AEI)/10.13039/501100011033 under Grant PID2020-118011GB-C21; in part by the Catalan Institution for Research and Advanced Studies (ICREA) through the Secretariat for Universities and Research of the Ministry of Business and Knowledge of the Government of Catalonia; in part by the European Social Fund; and in part by the National Research, Development and Innovation Fund of Hungary under Grant OTKA/ANN-135606, Grant OTKA/FK-135074, and Grant OTKA/FK-134604.
Document Type: Article
File Description: application/pdf
ISSN: 2169-3536
DOI: 10.1109/access.2025.3526751
Access URL: https://doaj.org/article/5550fbb44f13438687e6958ecc076db4
https://hdl.handle.net/2117/426782
https://doi.org/10.1109/access.2025.3526751
Rights: CC BY
Accession Number: edsair.doi.dedup.....d96ab9be5be8494191b6ed9bef049435
Database: OpenAIRE
Description
Abstract:Kubernetes is a widely used platform for deploying and managing containerized applications due to its efficient elastic capabilities. The Horizontal Pod Autoscaler (HPA) in Kubernetes independently adjusts the number of pods for each service, yet these services often operate in an interconnected manner. This study aims to understand the effects of autoscaling events on a graph of interconnected services. To achieve this, we apply control theory to model the HPA’s behavior. We analyze the stability of this model, perform numerical simulations, and deploy a real testbed to evaluate the performance. Our findings demonstrate that the control theory-based model accurately predicts the HPA’s behavior, ensuring system stability with CPU utilization meeting desired thresholds and no traffic loss after a transitional period. The model provides insights into optimizing resource scheduling and improving application performance in Kubernetes environments. Additionally, we extend our model to the whole service graph to understand how individual scaling decisions influence the complex graphs of cloud applications.<br />This work was supported in part by the Spanish I+D+i Project TowaRds fully AI-empowered NetwoRks, subproject A (TRAINER-A), funded by Ministerio de Ciencia e Innovación (MCIN)/Agencia Estatal de Investigación (AEI)/10.13039/501100011033 under Grant PID2020-118011GB-C21; in part by the Catalan Institution for Research and Advanced Studies (ICREA) through the Secretariat for Universities and Research of the Ministry of Business and Knowledge of the Government of Catalonia; in part by the European Social Fund; and in part by the National Research, Development and Innovation Fund of Hungary under Grant OTKA/ANN-135606, Grant OTKA/FK-135074, and Grant OTKA/FK-134604.
ISSN:21693536
DOI:10.1109/access.2025.3526751