Abstract

Cloud computing has emerged as a fundamental paradigm for large-scale distributed systems, continuously generating vast volumes of data, such as system logs and performance metrics. This paper presents a comprehensive literature review of anomaly detection techniques within cloud environments, categorized into Machine Learning and Deep Learning methodologies. We systematically analyze various models like clustering, classification, Autoencoders, and Recurrent Neural Networks (RNNs). A key contribution of this survey is the identification of a significant research gap: the inability of current industrial-level models to effectively handle high-dimensional telemetry data. Our investigation shows that a lack of integrated dimensionality reduction techniques in current frameworks creates scalability issues along with performance problems. Through the illumination of such important research gaps, this review serves as a guide on how more optimal anomaly detection solutions can be developed in an effort to improve cloud reliability.