Predictive Autoscaling for Containerized Applications Using Machine Learning

As containerization continues to gain prominence in modern application deployment, the need for efficient autoscaling mechanisms becomes paramount to ensure optimal resource utilization and adherence to service level objectives. This research paper proposes a Predictive Auto-scaling framework leveraging machine-learning(ML) techniques to anticipate and proactively adjust the resources allotted to containerized applications. The aim is to enhance the scalability and responsiveness of applications running in dynamic and cloud-native environments. An auto-scaling technique with burst awareness for containerized applications is proposed in this research. This innovative method optimizes container utilization while meeting the Quality-of-Service requirement by taking into account a mix of both vertical and horizontal skills. The proposed proactive autoscaling strategy addresses the challenges posed by unpredictable workloads and varying application demands, fostering adaptability to dynamic environments. By highlighting the significance of machine learning in improving the predictability and effectiveness of resource scaling for containerized systems, this study adds to new work of research on auto-scaling methodologies.