Advances and Predictions in Predictive Auto-Scaling and Maintenance Algorithms for Cloud Computing
Anantha Raman Rathinam, B. Santha Vathani, A. Komathi, J. Lenin, B. Bharathi, S. Murugan
Abstract
The use of sophisticated algorithms has radically altered cloud computing predictive auto-scaling and upkeep approaches. Recurrent Neural Networks (RNNs), the Prophet Algorithm, K-Means Clustering, and Seasonal Autoregressive Integrated Moving-Average (SARIMA) all play a role in improving cloud infrastructures, and their interactions are studied here. By capitalizing on their superiority in processing sequential data, RNNs can deduce accurate workload forecasts from past use patterns. Concurrently, the Prophet Algorithm records seasonal and annual patterns, which adds depth to forecasts. By grouping servers into clusters with similar consumption patterns, K-Means Clustering improves resource allocation efficiency and paves the way for precise auto-scaling. SARIMA models capture nuanced seasonal fluctuations, which lead to reliable demand forecasts. This explores the state-of-the-art and future directions of these techniques, illuminating their potential to revolutionize current approaches to cloud management. When these methods are combined, cloud service providers are better able to proactively scale their resources, hence reducing the likelihood of bottlenecks and outages. It foresees the development of these algorithms and their subsequent widespread use in a variety of fields outside of cloud computing, such as Internet of Things (IoT) networks and edge computing infrastructures.