Support for making decisions on managing computational resources of software services based on time series

The paper considers the task of estimating the current demand for computing resources by a micro-service based on monitoring data represented as time series, the results of which can then be used in decision-making on the allocation of computing resources. The demand is understood as the required intensity of consumption of the most in-demand computing resource, which ensures the processing of incoming tasks within a specified sampling window. A practical algorithm is proposed, which includes the following steps: selecting informative metrics using correlation analysis; synchronizing, normalizing, and converting metrics to a single discrete scale; cleaning telemetry from outliers, quantization, and missing values, and restoring them; detecting points of failure and segmenting them into operating modes; generating a baseline for consumption and restoring the demand series r(t). Two families of methods are considered for building the demand series, r(t). In the unsuper-vised setting, the specific resource consumption of a request is estimated based on the resource consumption and the counts of incoming/processed tasks, and exponential smoothing is introduced, with the smoothing parameters adapting to changes in modes. In the supervised setting, requests are divided into types/endpoints, and a regression problem is solved to estimate the costs of classes using NNLS and online RLS with a forgetting factor. The paper presents a methodology for experimental validation using a reproducible microservice testbed with controlled load scenarios, comparing the accuracy against the reference consumption, and estimating computational costs, making the results applicable for monitoring and scaling purposes.

1. Chen L., Yang X., Sun H., Yu X., Zhi K. Dependent task offloading in multi-access edge computing: A GCN augmented deep reinforcement learning approach. Computer Networks. 2025;269:111445. https://doi.org/10.1016/j.comnet.2025.111445

2. Ni Zh., Chen H., Xue H., Li Zh., Chen N., Yu J. Joint optimization of service placement, task offloading and resource allocation for dependent subtasks in hierarchical edge computing systems. Computer Networks. 2025;275:111896. https://doi.org/10.1016/j.comnet.2025.111896

3. Mytton D., Ashtine M.I. Sources of data center energy estimates: A comprehensive review. Joule. 2022;6(9):2032–2056. https://doi.org/10.1016/j.joule.2022.07.011

4. Ahmad H., Treude Ch., Wagner M., Szabo C. Towards resource-efficient reactive and proactive auto-scaling for microservice architectures. Journal of Systems and Software. 2025;225(9):112390. https://doi.org/10.1016/j.jss.2025.112390

5. Turin G., Borgarelli A., Donetti S., Damiani F., Johnsen E.B., Tapia Tarifa S.L. Predicting resource consumption of Kubernetes container systems using resource models. Journal of Systems and Software. 2023;203(4):111750. https://doi.org/10.1016/j.jss.2023.111750

6. Park J., Choi B., Lee Ch., Han D. Graph Neural Network-Based SLO-Aware Proactive Resource Autoscaling Framework for Microservices. IEEE/ACM Transactions on Networking. 2024;32(4):3331–3346. https://doi.org/10.1109/TNET.2024.3393427

7. Tawakuli A., Havers B., Gulisano V., Kaiser D., Engel Th. Survey: Time-series data preprocessing: A survey and an empirical analysis. Journal of Engineering Research. 2025;13(2):674–711. https://doi.org/10.1016/j.jer.2024.02.018

8. Blázquez-García A., Conde A., Mori U., Lozano J.A. A Review on Outlier/Anomaly Detection in Time Series Data. ACM Computing Surveys. 2021;54(3):56. https://doi.org/10.1145/3444690

9. Gupta M., Wadhvani R., Rasool A. Comprehensive analysis of change-point dynamics detection in time series data: A review. Expert Systems with Applications. 2024;248(2):123342. https://doi.org/10.1016/j.eswa.2024.123342

10. Grohmann J., Eismann S., Bauer A., et al. SARDE: A Framework for Continuous and Self-Adaptive Resource Demand Estimation. ACM Transactions on Autonomous and Adaptive Systems. 2021;15(2):6. https://doi.org/10.1145/3463369

11. Tzanettis I., Androna Ch.-M., Zafeiropoulos A., Fotopoulou E., Papavassiliou S. Data Fusion of Observability Signals for Assisting Orchestration of Distributed Applications. Sensors. 2022;22(5):2061. https://doi.org/10.3390/s22052061

12. Li B., Peng X., Xiang Q., et al. Enjoy your observability: An industrial survey of microservice tracing and analysis. Empirical Software Engineering. 2022;27(1):25. https://doi.org/10.1007/s10664-021-10063-9

13. Vorontsov Yu.A. Dimensional reduction of the set of monitoring metrics for assessing the state of computing systems. Scientific and Technical Volga region Bulletin. 2025;(5):218–223. (In Russ.).

14. Usman M., Ferlin S., Brunstrom A., Taheri J. A Survey on Observability of Distributed Edge & Container-Based Microservices. IEEE Access. 2022;10:86904–86919. https://doi.org/10.1109/ACCESS.2022.3193102

15. Cai B., Wang B., Yang M., Guo Q. AutoMan: Resource-efficient provisioning with tail latency guarantees for microservices. Future Generation Computer Systems. 2023;143(3):61–75. https://doi.org/10.1016/j.future.2023.01.014

16. Cabrieto J., Meers K., Schat E., et al. kcpRS: An R Package for Performing Kernel Change Point Detection on the Running Statistics of Multivariate Time Series. Behavior Research Methods. 2022;54(3):1092–1113. https://doi.org/10.3758/s13428-021-01603-8

17. Li Zh., Cao J. Automatic search intervals for the smoothing parameter in penalized splines. Statistics and Computing. 2023;33(1):1. https://doi.org/10.1007/s11222-022-10178-z

18. Filaretov G.F., Bouchaala Z. Development of Time Series Disorder Detection Algorithms Based on Nonparametric Hypothesis Testing Criteria. Bulletin of MPEI. 2021;(3):67–77. (In Russ.). https://doi.org/10.24160/1993-6982-2021-3-67-77

19. Aue A., Kirch C. The state of cumulative sum sequential changepoint testing 70 years after Page. Biometrika. 2024;111(2):367–391. https://doi.org/10.1093/biomet/asad079

20. Sholihat S.S., Indratno S.W., Mukhaiyar U. The role of parameters in Bayesian Online ChangePoint Detection: detecting early warning of mount Merapi eruptions. Heliyon. 2021;7(7):e07482. https://doi.org/10.1016/j.heliyon.2021.e07482

21. Loughman M., Barton S., Farrell R., Dooley J. Early Stopping Criterion for Recursive Least Squares Training of Behavioural Models. Wireless Personal Communications. 2022;126(2):1633–1648. https://doi.org/10.1007/s11277-022-09813-9

22. Zhu K., Yu Ch., Wan Y. Recursive Least Squares Identification With Variable-Direction Forgetting via Oblique Projection Decomposition. IEEE/CAA Journal of Automatica Sinica. 2022;9(3):547–555. https://doi.org/10.1109/JAS.2021.1004362