Designing ML systems that predict traffic spikes and scale cloud resources before demand rises, cutting idle costs
Author(s): Hema Vamsi Nikhil Katakam
Publication #: 2511027
Date of Publication: 06.10.2025
Country: United States
Pages: 1-7
Published In: Volume 11 Issue 5 October-2025
DOI: https://doi.org/10.62970/IJIRCT.v11.i5.2511027
Abstract
Cloud computing environments frequently encounter unpredictable surges during events such as tax-filing seasons or Black Friday sales. Traditional reactive auto-scaling mechanisms often respond too late, leading to degraded performance and inflated costs. This paper conceptually proposes an AI-driven predictive scaling framework that forecasts workloads using advanced machine learning models such as Temporal Convolutional Networks and Transformers. The design integrates monitoring, prediction, and adaptive decision modules to proactively adjust cloud resources. Simulated analyses indicate improved responsiveness and cost efficiency. The study remains conceptual, presenting an architecture for future empirical validation within real-world cloud orchestration systems.
Keywords: Auto-scaling, Predictive Analytics, Transformers, Cloud Optimization, Seasonal Workloads.
Download/View Count: 152
Share this Article