A comprehensive survey of Network Function Virtualization.

Computer Networks(2018)

引用 514|浏览205
暂无评分
摘要
Today’s networks are filled with a massive and ever-growing variety of network functions that coupled with proprietary devices, which leads to network ossification and difficulty in network management and service provision. Network Function Virtualization (NFV) is a promising paradigm to change such situation by decoupling network functions from the underlying dedicated hardware and realizing them in the form of software, which are referred to as Virtual Network Functions (VNFs). Such decoupling introduces many benefits which include reduction of Capital Expenditure (CAPEX) and Operation Expense (OPEX), improved flexibility of service provision, etc. In this paper, we intend to present a comprehensive survey on NFV, which starts from the introduction of NFV motivations. Then, we explain the main concepts of NFV in terms of terminology, standardization and history, and how NFV differs from traditional middle-box based network. After that, the standard NFV architecture is introduced using a bottom up approach, based on which the corresponding use cases and solutions are also illustrated. In addition, due to the decoupling of network functionalities and hardware, people’s attention is gradually shifted to the VNFs. Next, we provide an extensive and in-depth discussion on state-of-the-art VNF algorithms including VNF placement, scheduling, migration, chaining and multicast. Finally, to accelerate the NFV deployment and avoid pitfalls as far as possible, we survey the challenges faced by NFV and the trend for future directions. In particular, the challenges are discussed from bottom up, which include hardware design, VNF deployment, VNF life cycle control, service chaining, performance evaluation, policy enforcement, energy efficiency, reliability and security, and the future directions are discussed around the current trend towards network softwarization.
更多
查看译文
关键词
Network Function Virtualization,Virtual network function,Software defined networking,Algorithm,Network softwarization
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要