Tablets receive educational data from cloud servers in envisaged structure of Fatih (Movement of Enhancing Opportunities and Improving Technology) project. Constraints such as increasing educational data size, droping in internet connection speed and limited bandwidth cause increasing time and cost for data transfers. Employing school level proxy server in Fatih project ecosystem is promising approach to overcome these problems. It is critical to estimate idle time for synchronization between cloud and proxy server. In this study, two different idle time estimation approaches are compared. While former one is based on local system resources such as CPU and memory usage, latter one is based on measuring and clustering RTT values.