STORAGE SERVER DATABASE UTILIZATION FORECASTING USING HOLT-WINTERS AND ARIMA METHODS IN E-GOVERNMENT SYSTEM. STUDY AT KEMENKEU RI

Abstract

Storage is a storage medium which is an important part of the system infrastructure. Storage utilization is one of the variables in determining the needs and performance of a system. If the storage is lacking or depleted it will cause the system to be hampered and the possibility of interference occurring. There were eight complaints or reports of system disturbances recorded in the Ministry of Finance's ticketing service desk application caused by anomalies in storage or disk capacity. Things like this can be anticipated if we know the storage capacity requirements for the future. In compiling ICT capacity analysis at Pusintek there has been no use of past data to plan future storage requirements. The use of data mining algorithms can be used to obtain forecast values from storage utilization. The use of the ARIMA model and Holt-Winters Exponential Smoothing as a method used to predict inventory is considered to be a fairly accurate model. Both algorithms are used to compare which accuracy is better in predicting storage requirements by measuring RMSE. Research data was obtained from monthly utilization reports from January 2021 to December 2022. From the evaluation results it can be concluded that forecasting storage server database utilization using the Holt-Winters method is better than the ARIMA method with the RMSE results for the Holt-Winters method being 14332.661717740748 and the RMSE method ARIMA is 20498.977982137938. The results of this forecasting can be utilized in planning database server storage needs.