The Internet has been stunningly successful in modeling the way we access and exchange information in the modern world. However, its popularity has become the biggest impediment to its further growth. Due to its multi-provider nature, adopting a new architecture or modification of the existing one requires consensus among competing stakeholders. As a result, alterations to the Internet architecture have become restricted to simple incremental updates instead of radical changes by introducing and deploying new network technologies.
To introduce flexibility and diversity, separation of policy from mechanism is a well-examined principle in computing literature. Network virtualization reincarnates this very concept in the context of networking architecture. In this case, the role of traditional ISPs has been divided into two: infrastructure providers will just deploy and maintain physical networks, while service providers will deploy customized network protocols on virtual networks composed from multiple infrastructure providers’ resources, and offer end-to-end services to end users. By decoupling service providers from infrastructure providers, network virtualization introduces flexibility, promotes diversity, improves manageability, and decreases CAPEX & OPEX at the same time.
Even though it is hard to define network virtualization in a single sentence, the following one seems fair enough:
Network virtualization is a networking environment that allows multiple service providers to dynamically compose multiple heterogeneous virtual networks that co-exist together in isolation from each other, and to deploy customized end-to-end services on-the-fly as well as manage them on those virtual networks for the end-users by effectively sharing and utilizing underlying network resources leased from multiple infrastructure providers.