Elastic Auto-Scaling for Real-Time Anomaly Detection in Cloud Platforms
Wei Wu
Abstract
Cloud platforms increasingly demand effective solutions for real-time anomaly detection to ensure system stability and user satisfaction. We introduce Elastic Auto-Scaling, a framework that addresses this need by dynamically adjusting resource allocation according to both workload patterns and identified anomalies. By applying machine learning algorithms, Elastic Auto-Scaling can detect unusual behaviors in application metrics and performs automatic scaling in response, thereby reducing latency and optimizing performance under abnormal conditions. The inclusion of a forecasting mechanism enables the framework to predict resource requirements, facilitating preemptive adjustments that avoid potential issues. Comprehensive evaluations across diverse cloud scenarios reveal that Elastic Auto-Scaling outperforms traditional static scaling methods, significantly lowering detection times and minimizing resource waste. Its ability to adapt to various workloads positions it as a robust solution for enhancing cloud platform efficiency and improving the overall user experience.