Abstract:One of the key objectives of cloud computing systems is to meet the service level agreements (SLAs) under conditions of high resource utilization. Cloud service providers often need to design policies for resource sharing and performance optimization. As a result, being able to predict the performance and resource utilizations prior to implementing these policies is important to the dynamic provisioning of services by cloud providers. It is a significant and difficult challenge due to the fact that requests for resources often interact with each other in complex ways. Moreover, the dynamics of the cloud environment bring more problems to predicting the performance of a running query or workload. Hence, an accurate situation-aware model which can capture the complex interactions among resource requests is useful for addressing this challenge. To this end, we propose an efficient and highly accurate resource and performance prediction framework which takes into account the interactions among concurrently running resource requests for n-tier service systems. The proposed framework extends the Gaussian process and kernel canonical correlation analysis techniques and is able to dynamically adapt to variations in workload and physical resource usage. The proposed framework has been trained and evaluated extensively with a realistic multi-tier cloud application benchmark - the RUBiS benchmark system. The results demonstrate that the framework yields highly accurate performance and resource usage predictions, especially under high resource utilization conditions.

A Classification-Based Demand Trend Prediction Model in Cloud Computing

An Effective Classification-Based Framework for Predicting Cloud Capacity Demand in Cloud Services

A QoS Guarantee Framework for Cloud Services Based on Bayesian Prediction

A Comparison of Machine Learning Algorithms for Automatic Cloud Resource Scaling on a Multi-Tenant Platform

An Intelligent Analysis and Prediction Model for On-Demand Cloud Computing Systems

A Big Data Processing-Oriented Prediction Method of Cloud Computing Service Request

Resource and Performance Prediction at High Utilization for N-Tier Cloud-Based Service Systems.

PSO-SVR-Based Resource Demand Prediction in Cloud Computing

Cloud Analytics for Capacity Planning and Instant VM Provisioning

Performance and Resource Prediction at High Utilization for N-Tier Service Systems in Cloud an Experiment Driven Approach.

Survey on prediction models in cloud resource management schemes

CPU Load Prediction Using Support Vector Regression and Kalman Smoother for Cloud

Research on Resource Demand Prediction and Optimal Allocation Method in Cloud Computing Environment

A Study for Task Time Performance Dynamic Prediction Model in Cloud Resource Scheduling

Load Prediction-Based Automatic Scaling Cloud Computing

An Adaptive Prediction Approach Based on Workload Pattern Discrimination in the Cloud.

A Workload Prediction-Based Multi-Vm Provisioning Mechanism in Cloud Computing

On Dynamic Performance Estimation of Fault-Prone Infrastructure-as-a-Service Clouds.

Reducing Operational Costs Through Consolidation with Resource Prediction in the Cloud

Performance Evaluation and Cost Optimal Service Quality Prediction of Cloud Computing Systems in Unreliable Environments

Efficient Resources Provisioning Based on Load Forecasting in Cloud