Virtual Backup Server optimization on Data Centers using Neural Network

Abstract

The data center is a valuable asset in the organization because it keeps all important data of the organization, good data management is needed so that data is protected. To secure a data center, duplicate servers are generally created containing data backup from the data center. Every organization must provide high operational costs in providing special hardware (dedicated servers) for backup servers and high bandwidth for backup processes because duplicated data is all data from the data center. To minimize operational costs and high bandwidth usage, this study makes a virtual server backup system using the Neural Network. Using a virtual server does not require operational costs to buy and manage special hardware, and Neural Network uses the Iterative Dychotomizer version 3 (ID3) classification method and Backpropagation can solve the problem of backup data classification so that not all data in the data center is duplicated. The use of Neural Network using a combination of ID3 and Backpropagation classification methods, can accelerate the backup process and increase the accuracy of backup data when compared without Neural Network. The backup system research built is capable of producing backup processes in incremental backups with a time acceleration of up to 56.34% compared to the backup process without a Neural Network. In testing the accuracy of backup data shows that the backup process using the Neural Network has an accuracy level of 99.84% which is able to recognize duplicated data in accordance with the formation of a classification tree using ID3 and using Backpropagation for the learning process of duplicated data.