With the increasing number of business characteristics in the Internet of Vehicles (IoV) under the background of mobile edge computing (MEC), it is a challenge to realize the performance evaluation of server-to-end transmission in view of the complex modeling problem that needs to consider the service characteristics when designing the offload feedback strategy. In order to achieve a multi-dimensional comprehensive evaluation of performance, a cache scheduling strategy evaluation framework considering time-varying multi-type services is constructed based on queuing theory and Markov modulation service process. In order to adapt to various communication environments, the proposed framework can flexibly regulate business characteristics, double-ended processing rate and offload feedback strategy. Based on this framework, a offload feedback strategy based on statistical prediction method is proposed. Numerical results show that the transmission performance of the proposed strategy is 50% better than that of the traditional one, which also proves that the proposed framework can provide a reference for the design strategy under different communication environments and hardware configurations.