Review of AI-driven Cloud Optimization

Cloud automation is the key to realization a fully-optimized performance of modern cloud platforms while cloud resources utilization. Resource allocation efficiency is valuable. We are however faced with increasing pressure for computational resources. The Long Short-Term Memory (LSTM) algorithms have found a great use case in the dynamic resource allocation problem when the problem is solved by the proactive provisioning of resources based on historical usage patterns taking advantage of recurrent neural networks. Furthermore, the concern over quality-of-service delivery (QoS) and energy efficiency is now almost as challenging for cloud providers when implementing the utilization of cloud resources, especially in a dynamic setting. Deep Reinforcement Learning (DRL) allows to pursue that end by developing agents, which might guide the work of optimizing resource allocation and reducing energy expenses at the same time. It improves the result and adaptability of the applications running on clouds. Artificial intelligence, in its diverse form, for the instance machine learning and optimization algorithms, brings in a great influence in the areas of cloud operations, resource management, and security. Furthermore, the 3rd generation of FPGAs (Reconfigurable Digital Computing-In-Memory or ReDCIM processors) and the bitwise parallelism via-memory core multipliers also improve the efficiency of computing in cloud environments. Adopting these inputting methodologies is a consequence of cloud systems achieving top performance, high scalability, and low costing. Keywords - Cloud Automation, Long Short-Term Memory (LSTM) algorithms, Deep Reinforcement Learning (DRL), Reconfigurable Digital Computing-In-Memory (ReDCIM) processors, Resource sAllocation, Energy Efficiency.