Traffic-aware Dynamic Container Deployment on the Network Edge

Abstract

Multi-access Edge Computing (MEC) offers opportunities for improving the network performance. It can be utilized for improving web application. Every request that is destined for the cloud service providers can be processed at the edge. However, some web applications may have high fluctuation in terms of resource usage, for example e-commerce. The cloud server needs to be able to adapt with the traffic fluctuation by deploying the edge service automatically. We propose a technique for dynamically deploying the edge service by using statistical properties of the traffic, namely the moving average and moving standard deviation. These statistical properties are set as thresholds to decide whether to deploy the edge server or not. If the number of requests is higher than the maximum threshold, the edge server will be deployed. On the other hand, if it is lower than the minimum threshold, the edge server will be removed, if any. We build a testbed to evaluate our technique and the results show that it is feasible to use moving average and moving standard deviation for dynamic deployment.