Network virtualisation is a promising technique for a better future Internet by allowing for network resource sharing. However, resource sharing requires that virtual nodes and links be embedded onto substrate nodes and links (virtual network embedding), and thereafter the allocated resources dynamically managed throughout the lifetime of the virtual network (dynamic resource allocation). Since the constrained virtual network embedding problem is NP–Hard, many existing approaches are not only static, but also make simplifying assumptions, most of which would not apply in practical environments. This PhD research proposes improvements to both virtual network embedding and dynamic resource allocation. The objective is to achieve an efficient utilisation of physical network resources. To this end, we propose a path generation-based approach for a one-shot, unsplittable flow virtual network embedding, and a reinforcement learning-based dynamic allocation of substrate network resources.