Vehicular Ad hoc Network (VANET) is a dynamic network environment that supports the transmission of any type of data that varies with size and delay constraints. VANET also enables to aid safety message forwarding in case of emergency by means of data dissemination. The major problematic issues in routing and data dissemination are delay and broadcast storm. With the advancements in VANET technology not only normal traffic data transmission but also video streaming is transmitted while the delay constraint has to be very less for video streaming. This article addresses these two problems by designing a graph aware network management for emergency message dissemination by selecting an optimal relay using multi-objective shuffled shepherd optimization algorithm that computes degree, goodness, waiting time, transmission range and link utility. Graph is constructed with the link stability, signal strength, distance and speed between vehicles. The selected relay vehicle from the graph is responsible to disseminate the safety message, which reduces the broadcast storm. The use of reinforcement learning is able to select the most suitable low delay neighbors for video streaming. In routing, the normal traffic data transmission is performed which is tolerable to delay, so Type-2 fuzzy logic is used with dynamic belief entropy that defines routes with the computation of quality of forwarding, bandwidth, distance and hop counts. This proposed work is implemented on OMNeT++ simulator and the results show better performances than the existing work in terms of packet delivery ratio, end-to-end delay, duplicate packets, throughput, and frame loss.