Machine Learning-driven Dynamic Scaling Strategies for High Availability Systems

Abstract

High availability systems are essential in modern computing infrastructures to ensure uninterrupted service delivery and mitigate downtime risks. Dynamic scaling, the ability to adjust resources in real-time based on workload demands, plays a pivotal role in achieving high availability while optimizing resource utilization. Traditional scaling strategies often rely on static rules or manual interventions, which may not adapt effectively to fluctuating workloads and evolving system conditions.This abstract presents a novel approach leveraging machine learning (ML) techniques for dynamic scaling in high availability systems. By harnessing the power of ML algorithms, such as supervised learning, reinforcement learning, or deep learning, the proposed framework autonomously learns from historical data and current system state to make informed scaling decisions. This approach enables the system to adapt dynamically to workload variations, traffic spikes, and resource constraints, thereby enhancing scalability and resilience.