Predicting Resource Utilization for Cloud Workloads Using Machine Learning Techniques
Padma D. Adane, O. G. Kakde
Abstract
Padma D. Adane, O. G. Kakde
Abstract
Provisioning of resources in a cloud environment is a challenging issue. Under provisioning and over provisioning, both are detrimental for the overall performance of the system. There is a growing trend among researchers to use proactive provisioning approaches which anticipate the resource requirements in advance and prepare the system well to handle such real time demands. Although Proactive provisioning approaches are complex as compared to Reactive provisioning techniques, they provide an overall improved response time as the provisioning decisions are taken before the actual need of resources arise. The efficiency of such proactive provisioning techniques is dependent on the use of a predictive model that foresees the resource requirements. In this paper we have evaluated the performance of five popular Machine Learning Algorithms in predicting the CPU utilization of various server logs taken from the Parallel Workload Archive. The metrics used for evaluation are MAE- Mean Absolute Error and RMSE- Root Mean Squared Error.
OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Provisioning of resources in a cloud environment is a challenging issue. Under provisioning and over provisioning, both are detrimental for the overall performance of the system. There is a growing trend among researchers to use proactive provisioning approaches which anticipate the resource requirements in advance and prepare the system well to handle such real time demands. Although Proactive provisioning approaches are complex as compared to Reactive provisioning techniques, they provide an overall improved response time as the provisioning decisions are taken before the actual need of resources arise. The efficiency of such proactive provisioning techniques is dependent on the use of a predictive model that foresees the resource requirements. In this paper we have evaluated the performance of five popular Machine Learning Algorithms in predicting the CPU utilization of various server logs taken from the Parallel Workload Archive. The metrics used for evaluation are MAE- Mean Absolute Error and RMSE- Root Mean Squared Error.
Key concepts: Provisioning, Cloud computing, Computer science, Workload, Resource (disambiguation), Distributed computing, Mean squared error, Operating system