EUVIMEDEUVIMED entdecken
Crossref

METHOD FOR AUTOMATIC SCALING OF GOOGLE CLOUD SQL BASED ON LOAD METRICS

Marian Horban, Mykhailo Klymash, Roman Kolodiy

Infocommunication technologies and electronic engineering · 2026-11 · 6 (2) · 28-36

DOI: 10.23939/ictee2026.02.028 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Infocommunication technologies and electronic engineering, 2026-10. Verfasst von Marian Horban und 2 weiteren Autor:innen. Die Arbeit nennt 13 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Modern cloud applications often face unpredictable workload spikes that can significantly impact performance, system stability, and overall user experience. These spikes may be caused by irregular traffic patterns, batch processing jobs, or sudden increases in user activity, making static resource allocation inefficient and often economically suboptimal [1]. While Google Cloud SQL provides built-in capabilities for automatic storage scaling, it lacks a native mechanism for dynamic scaling of computational resources such as CPU and memory[2]. This limitation can lead to resource under-provisioning during peak loads or over-provisioning during idle periods. This paper presents a fully serverless, event-driven architecture for automatic scaling of Cloud SQL instances based on real-time CPU utilization metrics, implemented using Google Cloud Monitoring, Pub/Sub, Cloud Run, and the SQL Admin API [3]. The proposed system continuously monitors database performance metrics and evaluates them against predefined threshold values. When these thresholds are exceeded, alerting policies in Google Cloud Monitoring generate events that are published to a Pub/Sub topic. These events trigger a Cloud Run microservice responsible for executing scaling logic. The microservice interacts with the Cloud SQL Admin API to dynamically adjust instance configurations, including CPU and memory allocation, in response to current workload demands. Firestore is used as a state management layer to persist scaling decisions, maintain system state, and enforce a cooldown mechanism that prevents rapid oscillations between scaling actions, thereby improving system stability and avoiding unnecessary reconfiguration overhead. The study demonstrates that the integration of serverless and managed Google Cloud services enables the creation of a scalable, loosely coupled, and fault-tolerant system for database resource management. The proposed approach enables an automated controller-level response to workload changes and may improve cost efficiency under variable workload profiles by initiating compute-resource scaling only when predefined thresholds are met. This architecture is particularly suitable for small and medium-sized enterprises, where operational simplicity, cost optimization, and minimal infrastructure management overhead are critical requirements. Additionally, the solution can be extended to support more complex scaling strategies, including multi-metric decision models and cross-cloud adaptability.

Zitieren

APA

Horban, M., Klymash, M., & Kolodiy, R. (2026). METHOD FOR AUTOMATIC SCALING OF GOOGLE CLOUD SQL BASED ON LOAD METRICS. Infocommunication technologies and electronic engineering, 6(2), 28-36. https://doi.org/10.23939/ictee2026.02.028

Vancouver

Horban M, Klymash M, Kolodiy R. METHOD FOR AUTOMATIC SCALING OF GOOGLE CLOUD SQL BASED ON LOAD METRICS. Infocommunication technologies and electronic engineering. 2026;6(2):28-36. doi:10.23939/ictee2026.02.028

BibTeX
@article{Horban2026,
  author = {Horban, Marian and Klymash, Mykhailo and Kolodiy, Roman},
  title = {METHOD FOR AUTOMATIC SCALING OF GOOGLE CLOUD SQL BASED ON LOAD METRICS},
  journal = {Infocommunication technologies and electronic engineering},
  year = {2026},
  volume = {6},
  number = {2},
  pages = {28--36},
  doi = {10.23939/ictee2026.02.028}
}
Kostenlos starten →