ew 18: e1

Research Article

A Comparative study about Workload prediction from one time forecast with cyclic forecasts using ARIMA model for cloud environment

Download61 downloads
  • @ARTICLE{10.4108/eai.13-7-2018.163977,
        author={Yuvha Secaran R and Sathiyamoorthy E},
        title={A Comparative study about Workload prediction from one time forecast with cyclic forecasts using ARIMA model for cloud environment},
        journal={EAI Endorsed Transactions on Energy Web: Online First},
        volume={},
        number={},
        publisher={EAI},
        journal_a={EW},
        year={2020},
        month={4},
        keywords={Auto scaling, Time series, Workload prediction, environment, Energy, Cloud},
        doi={10.4108/eai.13-7-2018.163977}
    }
    
  • Yuvha Secaran R
    Sathiyamoorthy E
    Year: 2020
    A Comparative study about Workload prediction from one time forecast with cyclic forecasts using ARIMA model for cloud environment
    EW
    EAI
    DOI: 10.4108/eai.13-7-2018.163977
Yuvha Secaran R1,*, Sathiyamoorthy E1
  • 1: School Of Information Technology and Engineering, VIT University, Vellore, Tamilnadu, India
*Contact email: aspiringcoder9313@gmail.com

Abstract

Auto-scaling systems help provisioning resources on demand which helps tap into the elastic nature of the cloud. The applications hosted on the cloud tend to face workload surges which causes the response to be slow or denied. To tackle provisioning resources on demand there are reactive and proactive strategies in place. The topic of interest is the proactive strategies which uses a quantified metric as an input to provision resources before the demand arises. The quantified metric is the prediction obtained as a result of analysing the historical data of a application. This paper focuses using historical data of requests served by a web application to obtain a forecast value. The forecast value is the quantified metric which influences the scaling decisions. Conclusions are drawn about the accuracy of the metric based on prediction intervals along with the varied ways of forecast.