The next generation of mobile systems, the 5G, requires significant advances in system architecture design and development, as a result of the tremendous increase of mobile traffic, the omnipresence of mobile services and the ever-increasing number of connected devices. Several technologies, such as cloud computing, softwareization and virtualization have been identified to pave the road towards the efficient rollout and deployment of 5G. In order to address the strict service requirements, in terms of latency, reliability and throughput, they will have to be deployed at the radio access network edges. This paper discusses the importance and necessity for resource scaling in the mobile edge computing (MEC), as a promising technology approach for 5G. It also elaborates on the potential virtualization technologies, hypervisor-based and container-based, and the underlying resource scaling approaches for different relevant scenarios. The evaluation is performed on a laboratory demo platform capable of virtualizing several types of radio access systems, such as GSM, LTE, etc. The results clearly show that container-based virtualization provides the most efficient and robust resource scaling that does not significantly affect the communication performance of the mobile users.