Abstract:Cloud computing revolutionizes fast-changing technology. Companies' computational resource use is changing. Businesses can quickly adapt to changing market conditions and operational needs with cloud-based solutions' adaptability, scalability, and cost-efficiency. IT operations and service delivery have changed due to widespread computational resource access. Cloud computing efficiently allocates resources in cloud environments, making it crucial to this transformation. Resource allocation impacts efficiency, cost, performance, and SLAs. Users and providers can allocate cloud resources based on workloads using elasticity, scalability, and on-demand provisioning. IT economics and operational effectiveness have changed due to rapid and flexible resource allocation. Proactive versus reactive resource allocation is key to understanding cloud resource management challenges and opportunities. Reactive strategies allocate resources only when shortages or surpluses occur at demand. This responsive strategy often leads to inefficiencies like over- or under-allocation, which raises costs and lowers performance. Predictive analysis and workload forecasting predict resource needs in proactive resource allocation. Optimize resource use to avoid shortages and over-provisioning. Attention has been drawn to proactive predictive resource allocation. These methods predict resource needs using historical data, machine learning, and predictive analytics. Predictive strategies optimize resource allocation by considering future decisions. Reduced bottlenecks boost user satisfaction and lower operational costs. Matching resource distribution to workloads optimizes cloud resource management. Resource allocation prediction improves with deep learning. CNN, LSTM, and Transformer cloud resource forecasting algorithms are promising. New tools for accurate and flexible workload predictions have come from their ability to spot intricate patterns in historical data. This paper compares CNN, LSTM, and Transformer deep learning algorithms for cloud computing resource allocation forecasting. This study determines the best predictive accuracy and workload ada[1]ptability algorithm using Google Cluster Data (GCD). The study evaluates upgrading cloud computing resource allocation with the Transformer model. This study advances predictive resource allocation strategies, which can help cloud service providers and organizations improve resource utilization, cost-effectiveness, and performance in the face of rapid technological change.

Smartpick: Workload Prediction for Serverless-enabled Scalable Data Analytics Systems

ModiPick: SLA-aware Accuracy Optimization For Mobile Deep Inference

A Predictive Autoscaler for Elastic Batch Jobs

A hybrid cloud load balancing and host utilization prediction method using deep learning and optimization techniques

Intelligent Pooling: Proactive Resource Provisioning in Large-scale Cloud Service

CloudProphet: A Machine Learning-Based Performance Prediction for Public Clouds

Machine Learning for Performance Prediction of Spark Cloud Applications

A Deep Reinforcement Learning based Algorithm for Time and Cost Optimized Scaling of Serverless Applications

Optimizing serverless computing: A comparative analysis of multi-output regression models for predictive function invocations

Astrea: Auto-Serverless Analytics Towards Cost-Efficiency and QoS-Awareness

Optimal Resource Allocation for Serverless Queries

Optimizing simultaneous autoscaling for serverless cloud computing

Heterogeneity-aware Proactive Elastic Resource Allocation for Serverless Applications

Performance Analysis of Machine Learning Centered Workload Prediction Models for Cloud

Ditto: Efficient Serverless Analytics with Elastic Parallelism.

A proactive autoscaling and energy-efficient VM allocation framework using online multi-resource neural network for cloud data center

TR-Spark

Predictive Resource Allocation Strategies for Cloud Computing Environments Using Machine Learning

A simple and effective predictive resource scaling heuristic for large-scale cloud applications

Use of time-series based forecasting technique for balancing load and reducing consumption of energy in a cloud data center

End-to-end Optimization of Machine Learning Prediction Queries