This paper introduces the innovative Mobile Mist Computing (M2C) concept by addressing a reference architectural model for vehicular networks capable of supporting complex mobile services with stringent time‐space constraints. Cellular technology, thank to its coverage, data rates and latencies, is generally considered the key enabler to support real‐time vehicular applications. Moreover, their integration with the emerging Fog Computing (FC) paradigm can reduce the of the core network, thus minimising the overall service delay. In this scenario, vehicular mobility represents the most relevant challenge to guarantee service continuity, while providing Quality of Service (QoS), especially when Fog Servers (FSs) have limited resources. To this purpose, FC has been firstly integrated with the Software Defined Networking (SDN) and Network Functions Virtualization (NFV) concepts and, subsequently the dynamic replacement of vehicular services (ie, VNF migration) among FSs has been investigated to properly accommodate users mobility. Finally, an optimised strategy, called swapping migration, that can optimise both resources utilisation and outage rate, has been introduced. The proposed scheme has been validated by means of numerical simulations and compared with several benchmarks over realistic scenarios by pointing out latency and reliability as a function of the services request rate and the transmission capacity.