Practical Assessment of Large Language Models in Cloud Computing Using Real-World Data Applications

This study focuses on the application of Large Language Models (LLMs) within the modern cloud computing landscape, specifically addressing resource management and controlling proliferation. By leveraging Markov Decision Processes (MDPs) and Bayesian inference, the research achieved high accuracy in predictions and action efficiency. Data collected from leading cloud service providers (Azure, AWS, GCP, Oracle, and IBM) showed notable improvements in areas such as disk I/O performance, network and memory latency, and CPU processing capabilities. The findings demonstrate that LLMs outperform traditional models by optimizing job scheduling and reducing queue times. Bayesian inference played a key role in refining resource utilization estimates, while MDPs provided a framework for dynamic enhancements aimed at minimizing delays and boosting operational efficiency. LLMs are projected to play a pivotal role in the future management of cloud service cost, reliability, and performance. Future studies should explore object-based storage architecture, address the risks of unknown corruption, and tackle privacy concerns related to AI usage in cloud environments.