Network functions virtualization (NFV) is utilizing cloud computing and virtualization technologies to disaggregate the hardware and software of traditional networks and drive rapid development of new network services for telcos through a cost-effective and flexible approach. NFV is a key technology enabling wireless carriers to support demand for ever-increasing mobile broadband traffic and new enterprise use cases, such as network slicing.
Network functions virtualization (NFV) is the separation of network functions from the proprietary, physical hardware they run on by using virtual hardware abstraction. Virtual network functions (VNFs) are run in software and consist of virtual machines or containers, on generic, commodity hardware.
Dgtl Infra provides an in-depth overview of network functions virtualization (NFV), including how it works, the three main components of NFV architecture, and specific examples of how it is being implemented in the telecom industry. Additionally, we highlight the key benefits and challenges of NFV. Finally, Dgtl Infra reviews the key similarities and differences between network functions virtualization (NFV), software-defined networking (SDN), and network virtualization (NV).
What is Network Functions Virtualization (NFV)?
Network functions virtualization (NFV) is the separation of network functions from the proprietary, physical hardware (e.g., middleboxes) they run on by using virtual hardware abstraction. Through virtualization, network functions run in software on generic, commodity hardware (e.g., servers). These network functions include proxies, firewalls, intrusion detection system (IDS), deep packet inspection (DPI), network address translation (NAT), wide area network (WAN) accelerators, and load balancing.
NFV implements in software a virtual network function (VNF), which may consist of one or more virtual machines (VMs) or containers, running on industry-standard servers, switches, and storage devices – also known as commercial off-the-shelf (COTS) hardware. VMs or containers run this software, which accomplishes the same networking functions as traditional, physical hardware.
These virtualized resources can be located in cloud data centers, multi-tenant data centers, network nodes, and/or on-premises. Furthermore, NFV allows for the flexible deployment of VNFs across the four major cloud computing models: public cloud, private cloud, hybrid cloud, and multi-cloud. As such, VNFs can be moved to, or instantiated in, various locations in the network as required, without the need for installation of new hardware.
By utilizing cloud-native software, NFV eliminates the need for more expensive proprietary configurations or dedicated hardware. Instead, NFV allows software to be installed and run on large-scale deployments of more cost-efficient, industry-standard hardware, which pools computing resources across various network functions to further reduce costs.
As a solution, NFV focuses on benefitting telecommunications service providers and, particularly, wireless carriers such as Verizon, AT&T, BT, and Vodafone. NFV reduces a wireless carrier’s reliance on proprietary systems and hardware because individual network functions can be developed, provisioned, and upgraded independently of the underlying hardware.
By implementing network virtualization and automation, wireless carriers can make better use of their network infrastructure, which enhances service flexibility, elasticity, time-to-market, and the deployment of new network services – all while lowering costs.
Architecture of Network Functions Virtualization (NFV)
The three main components of network functions virtualization (NFV) architecture are: Virtual Network Functions (VNFs), NFV Infrastructure (NFVI), and NFV Management and Orchestration (NFV-MANO).
1) Virtual Network Functions (VNFs)
Network functions virtualization (NFV) is an approach to disaggregate and migrate a proprietary, physical network function (PNF) to a virtual network function (VNF). These VNFs run as software in a virtual machine (VM) or container, among a large number of top-of-rack (ToR) switches, in the public cloud data centers of Amazon Web Services (AWS), Microsoft Azure, and Google Cloud, as well as in private clouds. In turn, VNFs can scale on-demand to meet dynamic network performance or expansion demands.
With a NFV architecture, VNFs are deployed on-demand, reducing the provisioning time associated with traditional network hardware from months, to hours for virtual network components. Also, the need for on-site technical skills is reduced when VNFs are remotely deployed.
2) NFV Infrastructure (NFVI)
NFV infrastructure (NFVI) comprises the low-cost, standardized x86 computing hardware and software components that build the VM-based environment where VNFs are deployed. Specifically, NFVI delivers the virtual and physical resources, including compute, storage, and network, as well as virtualization software on which VNFs are deployed.
Through a hypervisor or a container management platform, NFVI provides the virtualization layer that sits above the hardware and abstracts hardware resources so they can be logically partitioned and provisioned to support VNFs. This virtualization layer allows network engineers to program all of the different segments of the virtual network and automate the provisioning of network resources.
NFVI is further evolving by using a microservices architecture and cloud-native network functions (CNFs), as well as Linux containers and the open-source Kubernetes orchestrator running on bare metal servers.
Cloud service providers (CSPs), such as Amazon Web Services (AWS) and Microsoft Azure, are using their platforms to offer NFVI solutions to telecommunications service providers. In particular, NFVI can be delivered centrally in large-scale data centers or deployed in a distributed manner, on-premises, through edge infrastructure and services like AWS Outposts and Microsoft’s Azure Stack Edge.
3) NFV Management and Orchestration (NFV-MANO)
NFV management and orchestration (NFV-MANO), also called MANO, is a framework delivered as orchestration software that is used for coordinating all resources in the cloud, which support the VNFs and the NFVI. Specifically, NFV-MANO is comprised of:
- Orchestration: NFV Orchestrator (NVFO) is a cloud-like feature, responsible for instantiating resources such as compute, memory, and storage to support the VM hosting the VNF
- Management: uses a VNF Manager (VNFM) to monitor and manage the lifecycle of a VNF running on a VM. Also, a virtual infrastructure manager (VIM) controls and manages the NFVI compute, storage, and network resources
Automating the management and orchestration of NFV onto virtualized cloud and edge infrastructure lets NFVI be more scalable and achieve better resource utilization.
Network Functions Virtualization (NFV) in Telecom
Historically, wireless carriers like Verizon have primarily relied on proprietary, hardware-centric solutions from a small number of incumbent network equipment vendors, such as Ericsson, Nokia, and Huawei. These rigid solutions have resulted in costly deployments and inefficient uses of network resources.
Given this backdrop, wireless carriers are pursuing cheaper and more efficient ways to connect subscribers and devices to their cellular networks, while meeting rapid increases in traffic demands, reducing latency, and offering new services. As such, many of the largest wireless carriers, including Verizon, AT&T, BT, and Vodafone are implementing NFV (which utilizes cloud computing and virtualization technologies) in their cellular networks.
Wireless carriers have transitioned from using traditional TDM-based voice and data networks, with proprietary network protocols, to standard IP networks. Increasingly, as a next step, these wireless carriers are migrating their compute by deploying virtual network functions (VNFs), through virtual machines (VMs), which run on industry-standard servers in the cloud.
NFV Market Size for Telecom
According to Omdia, the total wireless service provider NFV market for purchases of hardware, software, and services will grow from $17.2 billion in 2019 to $44.7 billion in 2025, representing a 17% compound annual growth rate (CAGR). Of this total, NFV software revenue is expected to grow from $9.0 billion in 2019 to $28 billion in 2025, representing a 21% CAGR.
Network functions virtualization (NFV) directly impacts i) cellular networks – spanning the core network to the radio access network (RAN) and ii) customer premises equipment (CPE). Moreover, the widespread implementation of NFV, through a cloud-native model, could impact wireless infrastructure, in particular, cell towers.
Wireless carriers are beginning to virtualize their core network and move their radio access network (RAN) to the cloud – known as Cloud RAN or C-RAN. This cloud-native approach requires network functions virtualization (NFV) and will bring about a step-function change in how cellular networks perform.
Furthermore, 5G is accelerating the adoption of NFV to satisfy rapidly growing traffic demand and the desire for real-time services, for use cases such as augmented reality (AR) and the Internet of Things (IoT).
The core network is the layer of the wireless carrier network which connects the radio access network (RAN) and the devices connected to it, to other network operators and service providers. In turn, the core network allows data to be transmitted to-and-from the internet or to-and-from other networks. The main function of the core network is to provide switching, routing, and transit for user traffic.
Initially, virtualization of the core network is occurring with multimedia messaging, the IP multimedia subsystem, and packet core components, including vEPC (virtual evolved packet core) and 5G Core (5GC). Importantly, the virtualized packet core allows for data connectivity in large-scale, private network or edge deployments, and permits the cost-effective transfer of data packets across wireless networks.
Virtualization of the core network is occurring as follows:
- vMessaging: enables wireless carriers to offer rich communication services (RCS), for next-generation messaging and video, without an OTT (over-the-top) application installed
- vIMS (virtual IP multimedia subsystem): provides call control for services such as voice, video, rich messaging and thus enables software for the core network and mobile services
- vEPC (virtual evolved packet core): organized in independent slices of the control, user, and management planes, vEPC is free of the architectural restrictions posed by traditional, physical node-based packet cores. This allows the virtualized user plane of gateways to be deployed in different locations and run on industry-standard hardware (i.e., servers). vEPC has been implemented in 4G/LTE networks of all sizes and is natively extensible to 5G standards with its service based architecture (SBA)
- 5G Core (5GC): supports a programmable data path that handles massive throughput (higher Gbps) with a reduced hardware footprint. Built on industry-standard hardware and open-source software platforms, 5G Core solutions employ cloud-native frameworks like using a microservices architecture and requires container-based deployment with Kubernetes orchestration. Wireless carriers can implement the shift from a single, rigid core network, toward a core that provides different logical networks, or “slices”, for different traffic requirements to support new use cases (e.g., IoT)
Radio Access Network (RAN)
The radio access network (RAN) is a wireless variant of the access network and refers to a cellular network such as 4G or 5G. For wireless carriers, the RAN requires a considerable amount of expenditure when building-out each new generation of wireless telecommunications technology, with 5G continuing this trend.
The 5G RAN is being supported by distributed compute, storage, and network resources at the infrastructure edge because it utilizes a combination of network functions virtualization (NFV), software-defined networking (SDN), and Cloud RAN (C-RAN) technologies.
Virtualization of the radio access network (RAN) is occurring as follows:
- OpenRAN and vRAN: an OpenRAN-based virtualized Radio Access Network (vRAN) delivers ubiquitous wireless connectivity, mobility, and edge services and allows all mobile devices – from cell phones to connected cars – and subscribers to connect to a network to access its services. OpenRAN-based vRAN solutions support 3G, 4G, and 5G networks with multiple functional splits, enabling disaggregation and cloud operation. This solution allows the RAN to be elastic, scale, and adapt based on usage and coverage. In turn, this RAN flexibility unlocks expanded and more convenient network location choices for the baseband processing on industry-standard hardware. Virtualized baseband solutions can be located in points-of-presence (PoPs) or aggregation data centers, serving clusters of cell sites
- Multi-Access Edge Computing (MEC): moves the computing of traffic and services from a centralized cloud to the edge of the network and thus closer to the subscriber or device. This allows for connectivity and data processing to occur at the edge of wireless carrier networks, such that processing tasks can be more closely located to subscribers or devices, thereby reducing latency and cost
- Private Networks: solution that offers a dedicated, secure wireless network to enterprise customers, built on top of OpenRAN-based vRAN, MEC, and software for the core network. Enterprises and industries can use 4G/LTE and 5G networks to replace or augment existing wired and wireless local area networks (LANs). These solutions can be offered in two models: i) small footprint rack-mountable servers deployed on-premises or ii) a hosted cloud delivered by a provider, on-premises or off-site, in a colocation data center
Ultimately, virtualization of the radio access network (RAN) will enable multiple wireless carriers to share the same physical resources so that their coverage and resource utilization are both increased.
Customer Premises Equipment (CPE)
Network functions virtualization (NFV) centralizes network functions onto industry-standard hardware in the cloud for devices located at end user premises by virtualizing customer premises equipment (CPE). By running software-based functions on shared infrastructure, this virtualized CPE can deliver superior speed, agility, operational simplicity, and cost reductions.
With NFV, the CPE can simply act as a forwarding device while virtual network functions (VNFs) can be run in a cloud data center. More specifically, virtualization and cloud-native CPE solutions include: virtual customer premises equipment (vCPE) and virtual enterprise customer premises equipment (vE-CPE), which is also known as universal customer premises equipment (uCPE).
Importantly, these virtualized CPE solutions can deliver multiple functions simultaneously, making them easier to monetize than traditional CPE, which has bundled proprietary hardware and software.
Example – SD-WAN (Software-Defined Wide Area Network)
Telecommunications service providers are utilizing uCPE for the deployment of SD-WAN (software-defined wide area network) solutions, which connect users and applications on enterprise networks across many locations (e.g., branch offices). By integrating NFV infrastructure (NFVI) with SD-WAN, uCPE servers can host virtual network functions (VNFs), consisting of virtual machines (VMs) or containers that are managed via a centralized orchestration system, such as Kubernetes.
Wireless infrastructure owners and operators including Crown Castle, SBA Communications, and IHS Towers have each explicitly cited network functions virtualization (NFV) as a risk related to their business. Particularly, improvements in the efficiency, architecture, and design of wireless networks – such as placing more compute closer to cell towers and small cells – could reduce demand for this wireless infrastructure.
As an example, NFV promotes network sharing and joint development by a cell tower owner’s wireless carrier customers (e.g., Verizon, AT&T, and T-Mobile). This pooling of resources could reduce the need for new wireless infrastructure, if wireless carriers utilize shared equipment rather than deploy new equipment. Moreover, NFV could result in the decommissioning of equipment on certain cell towers because portions of a wireless carrier’s network may become redundant.
What are the Benefits and Challenges of NFV?
Below we highlight the benefits and challenges of NFV from the perspective of telecommunications service providers, such as wireless carriers.
Benefits of Network Functions Virtualization (NFV)
The benefits of network functions virtualization (NFV) are a lower total cost of ownership, flexibility, elasticity, time-to-market, and its open approach.
1) Lower Total Cost of Ownership
Network functions virtualization (NFV) significantly reduces the capital expenditures (CapEx) and operational expenditures (OpEx) of a network in three primary ways:
- Equipment Costs: wireless carriers can migrate away from expensive proprietary, physical hardware components which take up significant space, by delivering software-driven capabilities. Additionally, because NFV runs on virtual machines (VMs) instead of physical machines, fewer hardware components are necessary and operational costs are lower
- Energy Consumption: reduced power consumption through consolidating equipment and leveraging the economies of scale of the cloud service providers (CSPs)
- Personnel: virtualization and automation make adding or removing new VNFs easily manageable by the wireless carrier, without requiring the physical presence of technicians on-site or having enterprise customers involved
As a case study, Vodafone, one of the largest wireless carriers in the world, completed the roll-out of NFV infrastructure (NFVI) from VMware across all its European business and 21 markets in total. Based on Vodafone’s internal analysis, the company saw gains in productivity as it brought network functions online around 40% more quickly, and cost savings of up to 55%.
Network functions virtualization (NFV), which deploys virtual network functions (VNFs) through software, is much more flexible than proprietary hardware running a physical network function (PNF). VNFs can be deployed on-demand or moved to various locations in the network as required, without the need for installation of new hardware. This flexibility is beneficial for wireless carriers because resources – such as compute, storage, and network – can be instantiated as-and-when needed, to support the VM hosting the VNF, which, in turn, allows for high resource utilization.
Network functions virtualization (NFV) combines the use of virtualization with the elasticity that comes from cloud data centers filled with industry-standard servers, switches, and storage devices. By being cloud-native, NFV utilizes elastic cloud services, which are built to be easily and automatically resized. In turn, telecommunications service providers can automatically add or remove resources, based on pre-defined policies, to meet changing network traffic demands and “spin up” or destroy virtual network functions (VNFs).
In contrast, telecommunications service providers operating physical hardware at their central office do not benefit from the elasticity of the cloud. Instead, they would be hindered by the need to have available and pre-configured computing resources at all times to handle any potential changes in traffic demand on their network.
Historically, launching new physical network functions (PNFs) and services was time consuming, with wireless carriers having to allocate space and power, along with needing to integrate another hardware device into a very large network. Instead, with network functions virtualization (NFV), VNFs and new services can be provisioned dynamically, without installing new hardware. As such, deploying network components with NFV takes hours, unlike traditional networking where it takes months.
Also, separating network functions software from proprietary, physical hardware, facilitates a faster pace of innovation and a shorter development cycle, which results in a quicker time-to-market for new services.
Network functions virtualization (NFV) is open and vendor agnostic. NFV disaggregates the network, taking proprietary hardware that houses captive software and breaks it apart into open, standardized systems.
For telecommunications service providers, this results in a diversified hardware supply chain and no vendor lock-in. At the same time, these service providers have the ability to choose virtual network functions (VNFs) from different vendors based on their requirements.
Also, NFV facilitates easier integration and faster deployment of VNFs into open-source orchestration systems, such as Kubernetes, which provide management of the virtual infrastructure.
Challenges of Network Functions Virtualization (NFV)
The challenges of network functions virtualization (NFV) are scalability, interoperability with application programming interfaces (APIs), performance, security, and resiliency. Below are further details on the potential disadvantages of NFV:
Network functions virtualization (NFV) seeks to deliver scalable services that will automatically grow in capacity to seamlessly meet any changes in demand. However, in very large networks, the issue of scale arises because it can be difficult for NFV architectures to grow to support millions of subscribers and devices across large geographic areas (i.e., nationwide). At the same time, NFV architectures have to cope with increasing traffic volume and a greater number of virtual network functions (VNFs) being deployed.
2) Interoperability with APIs
Physical network function (PNFs) remain prevalent in large-scale wireless carrier networks, as network functions virtualization (NFV) is being implemented slowly and incrementally in the core network and radio access network (RAN) of wireless carriers. As such, wireless carrier networks still rely on many proprietary, physical hardware devices that are complex to operate and difficult to integrate because of their vendor-specific application programming interfaces (APIs). In turn, it is challenging to connect these APIs with open-source orchestration systems, such as Kubernetes.
Wireless carriers have strict requirements for network performance, which are usually contractual, in the form of service level agreements (SLAs). For example, an SLA may specify the average latency, bandwidth, and the availability (uptime) for all the services that a network provides to one wireless carrier customer (e.g., Verizon).
To support SLA compliance, network functions virtualization (NFV) needs to be able to monitor virtual network functions (VNFs) for each customer and dynamically adapt compute and network resources. Optimizing VNF performance and automatically scaling VNF resource allocation with workloads is a challenge for NFV.
Network functions virtualization (NFV) and its virtual network components are vulnerable to different risks from security attacks and malware than those faced by physical hardware housed in a data center. For example, malware is more difficult to isolate and contain in a virtual environment than it is between hardware components, which are able to be physically separated.
In traditional networks, where both control and data packets are transmitted on the same connection, the control and data information are equally impacted when a failure occurs. In a NFV deployment, individual component failures can occur in both software and hardware, impacting resiliency.
Network Functions Virtualization (NFV) vs Software-Defined Networking (SDN)
Network functions virtualization (NFV) and software-defined networking (SDN) are both independent approaches to networking that serve different goals. However, they are also complementary and overlapping in a number of different ways.
Similarities between NFV and SDN
Network functions virtualization (NFV) and software-defined networking (SDN) are both software-based approaches to networking that rely on virtualization technology to function and benefit from automation. Both NFV and SDN use network abstraction, commodity hardware, and software to support more efficient and programmable network services:
- NFV: abstracts networking functions from the proprietary, physical hardware on which they run and, instead, through virtualization, network functions can run in software on generic, commodity hardware
- SDN: decouples network control and packet forwarding functions from closed and proprietary physical hardware and, instead, utilizes programmable commodity hardware and standards-based software to control packet forwarding
NFV and SDN’s use of cloud computing and virtualization technologies is fundamentally changing the roles of data centers, networks, wireless carriers, and internet service providers (ISPs). Particularly, these approaches are resulting in a lower total cost of ownership, flexibility, automation, and an open ecosystem for network providers.
Differences between NFV and SDN
Network functions virtualization (NFV) can be implemented independently of software-defined networking (SDN) because it is possible to virtualize network functions without using SDN approaches. At the same time, SDN may be used for many purposes unrelated to NFV.
While NFV separates networking services from dedicated hardware, SDN separates the network control functions, such as routers and switches, from packet forwarding functions.
NFV aims to reduce the cost and time to provide network functions that support the delivery of a service, but it does not introduce changes to existing protocols. In contrast, SDN’s scope is much broader, as it controls and manages a series of network objects that could contribute to a service, by decoupling and centralizing the network intelligence from the packet forwarding process.
When Should NFV be used with SDN?
Network functions virtualization (NFV) is commonly implemented in conjunction with software-defined networking (SDN) when there are many physical network elements that need to be virtualized. SDN can assist NFV by refining the process of controlling data packet routing through a centralized server, improving visibility and control.
More precisely, SDN is typically used for dynamically establishing a connection between VNFs, while NFV can handle the complex task of managing, monitoring, and orchestrating a large number of VNFs and their service chains.
NFV moves services to a virtual environment but does not include policies to automate the environment. Together, SDN’s centralized management function can forward data packets from one network device to another, while NFV allows routing control functions to be placed on a virtual machine (VM) or a container running on industry-standard servers, switches, and storage devices.
READ MORE: Software-Defined Networking (SDN) Explained
Example – Network Slicing
Network functions virtualization (NFV) and software-defined networking (SDN) are designed to support, configure, and implement new services, such as network slicing. Network slicing allows multiple virtual networks to be created on top of a common shared physical infrastructure. Each virtual network “slice” is a logical network that provides specific capabilities and performance characteristics in order to serve a defined business purpose of a customer.
Network slicing enables different types of 5G use cases, depending on the latency, speed, and availability (uptime) characteristics that are required. For example, specific network slices can be created for different purposes:
- Mission-Critical Communication: guaranteed voice and data connectivity between first responders and emergency control requires high bandwidth and uptime in the network, as well as prioritized capacity
- Mobile Gaming: low latency is needed for lag-free online gaming on a smartphone
Network Functions Virtualization (NFV) vs Network Virtualization (NV)
Network functions virtualization (NFV) adds virtual functions to the physical network, whereas network virtualization (NV) adds virtual tunnels to the physical network. The main differences between NFV and NV are as follows:
- Network Functions Virtualization (NFV): virtualizes OSI layer 4 through 7 functions. Specifically, these network functions include proxies, firewalls, intrusion detection system (IDS), deep packet inspection (DPI), network address translation (NAT), wide area network (WAN) accelerators, and load balancing
- Network Virtualization (NV): creates an overlay of the physical network. Instead of connecting two different endpoints with physical cabling in a network, network virtualization creates tunnels through the existing network. This is optimal for providing connectivity between virtual machines (VMs)