The spread of the internet applications has affected the vehicle connectivity requirements where connected vehicles can communicate with other systems to exchange data outside the vehicles. The continuous connectivity of smart vehicles is becoming very important for supporting various user applications. The challenges arise from the moving nature of vehicles on the road and the ability of providing a stable network connection due to areas where the network coverage is not available. The use of predictive multi-homed switching can eliminate the effect of network coverage holes. A switching decision can be supported by the intelligence of a connectivity gateway that analyzes network connectivity metrics collected from the vehicles such as round time delay and packet loss rate. In this paper we study the use of predictive multi-homed switching and the associated connectivity gateway in order to evaluate the improvement on the service connectivity of the vehicles. The gateway provides information ahead of time to prepare the vehicle to utilize alternative connectivity methods on different areas along the path of the vehicle. The system's reaction to the elimination of network coverage holes is assessed where we show the improvement in the continued connectivity key performance indicators.