What is network functions virtualization (NFV)?

A traditional network appliance approach requires each network function such as load balancers, firewalls, gateways and routers to run on dedicated hardware, which can be expensive and difficult to scale.

Making upgrades to dedicated hardware can take months or require purchasing new hardware compatible with new technology. Decoupling these network functions from hardware appliances enables service providers to greatly increase the speed of new service deployment while reducing the need for physical devices.

NFV architecture uses virtualization and VMs to create an agile network that is scalable, customizable and manageable through a single pane of glass. A centralized control panel enables network operators to automate the provisioning and orchestration of network resources and quickly respond to changing traffic patterns and network demands.

NFV serves the growing number of enterprises seeking to retain control of their network infrastructure while migrating from physical hardware to virtualized and cloud computing resources. For this reason, Forbes lists NFV in the top five technologies evolving telecom services alongside artificial intelligence (AI) and machine learning (ML), edge computing, APIs and computer vision.¹

NFV architecture establishes a foundation, process and strategy for the virtualization of network functions. The European Telecommunications Standards Institute (ETSI) Industry Specification Group for Network Function Virtualization (often referred to as ETSI ISG NFV) produced a white paper laying out the initial open source framework for NFV. Other organizations have participated in the development of NFV, but the basic architecture remains the same.

NFV architecture consists of three layers:

VNFs are the services previously running on physical hardware. Routing, firewalls, IP configuration and intrusion detection systems, SD-WAN systems and file sharing programs are common types of virtualized network functions.

When virtualized, these services can link together in a process called “service chaining.” Service chaining helps network operators automate the provisioning of resources for every service on the network. Having a centralized view of all functions gives operators enhanced network control and enables them to steer traffic and workloads to available servers, which cuts the risk of service outages.

NFV infrastructure consists of the servers, storage, switches and compute resources needed to create NFV environments. To abstract network functions from physical hardware, network operators create a virtualization layer, by using software called a hypervisor. A hypervisor, or virtual machine monitor (VMM), creates a software layer capable of segmenting multiple virtual machines from a single physical machine. These virtual machines can run alongside each other on their own operating system. NFVI offers connectivity to create a unified network out of multiple physical and virtual machines.

NFV MANO is the basic framework for managing the deployment, provisioning, monitoring and performance of virtualized network functions. NFV MANO also creates an interface for NFVI to communicate and interact with existing operating support systems (OSS) and business support systems (BSS).

MANO is divided into three subsections:

Uses virtualization technologies to deploy new network functions and provision resources to existing VNFs. NFV orchestration also serves to authenticate NFVI resource requests.

Optimizes the lifecycle of software, virtual resources and physical network. VIM instances can manage multiple NFVI resources or specialize on a specific aspect as needed. A VIM keeps a record of virtual and physical resources allowing network operators to maintain operations and deploy new services.

Standardizes virtual network functions and increases the interoperability of software-defined network (SDN) features. VNF management includes the instantiation, or creation of, instances, scaling, upgrading and termination of virtualized network functions.

To capitalize on the benefits of network functions virtualization, IT teams must recognize and address some challenges, particularly the visibility and security concerns that can accompany NFV. 

NFV environments often require more complex monitoring tools to monitor the various virtual machines, functions and traffic moving through the network. Virtualized functions are also more open to cyberattacks and malware than physical hardware stored in a data center and need to be protected in different ways. Enterprises seeking to deploy NFV should pair this transition with robust, NFV-specific monitoring and security practices to protect their data and infrastructure.

Software-defined networking (SDN) and network function virtualization share common elements but serve distinct functions and use cases. Both functions are software-defined approaches that are based on creating a virtualized layer above a physical network that is used to make networks more flexible.

However, SDN is focused on data centers, while NFV is geared toward wide-area networks (WANs) and network service providers and operators. While NFV virtualizes network functions and is used to reduce the need for physical devices (increasing agility and reducing costs), SDN helps organizations centralize network management and improve route network traffic.

SDN achieves this by decoupling the control plane—which organizes and sets controls for how data packets are routed throughout a network—from the underlying data plane, which is the engine that moves the data packets.

This centralization enables more precise management of network resources based on organization-specific policies and more efficient use of automated provisioning. Network operators can deploy automation tools and dynamically load balance and provision resources based on real-time conditions, which reduces latency and boosts overall service delivery.

Virtual network functions can be deployed across an SDN ecosystem. Used together, SDN and NFV help create agile, flexible networks capable of managing complex virtual environments.