Due to the powerful storage and computing capability, wireless communication capability and abundant energy supply of vehicle nodes and roadside facilities, the vehicular ad-hoc network (VANET) can detect change of vehicle driving environment, evaluate dangerous road conditions and warn drivers, such as warning of accident sites, collision prevention at intersections, predict the reaction time of drivers, so as to provide technical support for safe driving and driving experience. However, data transmission in a VANET faces multiple challenges, such as unstable wireless channel quality, time-varying network topology, short lifetime of wireless links, limited bandwidth and large communication load. Therefore, how to transmit data timely and reliably to the "urban brain" has become a issue. Taking storage cost, packet loss penalty and transmission reward into account, In this paper, a hybrid flow scheduling strategy for VANETs is first proposed, which allocates transmission resources for data flows with different priorities; then, combined with link status, a data transmission strategy for joint optimization of hybrid flow scheduling and path selection is introduced to meet the requirement on delay for real-time data and improve reliability for non real-time at the same time. Simulation results and analysis demonstrate the performance of proposed strategies.