Automotive systems will more and more rely on cloud-provisioned services as a means to handle the increased computational demands of intelligent vehicle functions and to support the driver in innovative ways. Especially challenging in this context is the dynamic nature of vehicles and the inconsistent state of their environment, causing a rapid change of driving situations which need to be handled in different ways. As a reaction to this, we provide a conceptual framework which allows for the definition of operational situations and operational strategies as a way to respond to these situations. Responses may involve automatically triggered reactive measures, rule-based strategies, and AI-controlled measures. We furthermore provide an exemplary technical solution upon which such framework may be realized. We propose to split vehicular functionality into isolated services which may be replicated individually and moved between the vehicle's onboard system and the cloud at run-time, depending on situational demand. Connectivity between vehicular and cloud-based services is provided by virtual overlay networks and a publish-subscribe middleware. Finally, we provide benchmarks as a means to evaluate the proposed technical realization. We find that our approach is indeed feasible and performs well. However, there is evidence that the employed technologies do not meet the high demands of the automotive domain, yet. Nevertheless, the fundamental mechanisms are successfully demonstrated.