Dynamic Fault-Tolerance and Mobility Provisioning for Services on Mobile Cloud Platforms

Abstract

We present the implementation of a platform for providing mobility and reliability to computational tasks that are executed in the access cloud. The platform is primarily aimed at providing an experimental tool for further studying of sophisticated techniques for task scheduling, placement and migration policies. In its current form, the platform supports streaming services but can be extended with support for synchronisation between computational tasks. The implementation builds on the Calvin platform developed by Ericsson. Being designed for implementation of services for IoT applications and written in Python, the platform provides excellent flexibility and is easy to extend at the cost of limitations to the task sizes that can be supported with stringent migration time limits. In this paper, we present the platform and its main functions and present experimental results which show how the platform performs as the task sizes increase.