Network Function Virtualization Definition
Network function virtualization or NFV is a concept in network architecture that decouples hardware and and network functions using virtualization technologies. By virtualizing entire categories of network node functions into modular units, NFV achieves greater scalability in communication and computing services.
NFV utilizes traditional server-virtualization methods like those deployed in enterprise IT, but it is unique. Custom hardware appliances for each network function are not necessary for a virtualized network function (VNF). Instead, one or multiple virtual machines (VMs) deploying distinct processes and software on top of switches and storage devices, typical high-volume servers, or cloud computing infrastructure can comprise a VNF.
Network function virtualization examples include virtualized load balancers, session border controllers, firewalls, WAN accelerators, intrusion detection devices, and more. Administrators may deploy any of these to deliver network services or protect a network without the typical complexity and cost of acquiring and installing physical units.
What is Network Function Virtualization?
Network function virtualization unlinks network services from proprietary hardware appliances, enabling them instead to run in virtual machines as software. Admins can virtualize standard compute, storage, and network function resources and place them on commercial off-the-shelf (COTS) hardware such as x86 servers. Giving available x86 server resources to the VMs keeps network services independent of hardware.
In this way, network function virtualization allows multiple VNFs to run on just one server and scale to consume the free resources that remain. This virtualization of infrastructure also typically results in more efficient use of data center resources. Both outside networks and within the data center, NFV can also virtualize the control plane and data plane.
NFV Performance, Background, and History
In the telecommunication industry, product development has followed rigorous standards for protocol adherence, stability, and quality. However, standards for hardware development led to slow development, long product cycles, and reliance on proprietary hardware. Public internet communications services such as Google Talk, Netflix, and Skype along with consumer demand drove changes to this status quo.
In October 2012, a working group on network function virtualization published a white paper on OpenFlow and software-defined networking (SDN). This paper from the group, part of the European Telecommunications Standards Institute (ETSI), began the movement toward network function virtualization.
To realize the enhanced benefits of virtualization, NFV equipment vendors continue to improve IT virtualization technology to achieve scalability, high availability, improved network management capabilities, and more effective performance. Efficient implementation of carrier-grade features is critical to minimizing the total cost of ownership (TCO). This level of efficiency demands that NFV solutions achieve five-nines availability (99.999%) by effectively using redundant resources.
Virtualization is changing how administrators specify, measure, and achieve availability in NFV solutions. As VNFs replace traditional equipment dedicated by function, a layered, end-to-end, service-based approach from network function virtualization companies is superseding approaches limited by equipment availability. The broken link between specific hardware or equipment and functions allows VNF services to define availability instead.
Different types of NFV functions each come with their own set of user expectations for service availability, and NFV technologies can virtualize a wide range of function types. Therefore, NFV solutions should support a broad spectrum of fault tolerance options. This level of flexibility should allow NFV solution providers to meet any VNF availability requirement.
Network Function Virtualization Infrastructure
NFV allows for a flexible, open architecture with a wide range of deployment options and NFV solutions. The typical NFV architectural framework is made up of three distinct layers:
Virtualized network functions (VNFs)
VNFs are software implementations of network functions, such as load balancing, firewall, IP multimedia subsystems, mobile core, routing, security, or video. Virtualized network functions can be deployed on a network functions virtualization infrastructure (NFVi).
Network functions virtualization infrastructure (NFVi)
NFVi or NFV infrastructure is the universe of all software and hardware components that comprise the deployment environment for NFVs. The NFV infrastructure can stretch across multiple locations, and the networking equipment that connects those locations is part of the NFVi.
Network functions virtualization management and orchestration architectural framework (NFV-MANO)
NFV-MANO includes all functional blocks, the data repositories they use, and the interfaces and reference points through which they exchange information. Information exchange is critical for VNF and NVVi orchestration and management purposes. Similarly, NFV microservices build out highly complex functions solely in the cloud using modular, distributed software components.
NFV is a building block for both the NFVi and the NFV-MANO. The NFV platform serves as virtualization software in both roles. In its NFVi role, it also provides both physical and virtual storage and processing resources. In NFV-MANO mode, the NFV platform operates both virtualization software and NFVi and VNF managers on a hardware controller. In addition, the NFV system also provides important public carrier network requirements, such as carrier-grade monitoring and management features for platform components and tools for security and recovery from failures.
Advantages and Disadvantages of Network Function Virtualization
Network function virtualization can better scale and adjust resources available to applications and services, shorten the time-to-market for new or updated products, and save money. More details about the advantages of network function virtualization include:
NFV in IoT or another form of virtualized infrastructure enables increased workload capacity with less—less power consumption, a smaller data center footprint, and similar or reduced cooling requirements. Fewer servers can do the same amount of work because just one server can run multiple VNFs at once.
As network demand fluctuates, software can update organizational infrastructure instead of physical appliance updates to data centers and the network. Network function virtualization also allows multiple functions to run on a single server, eliminating proprietary physical hardware, consolidating resources, and reducing costs.
Reduced vendor lock-in
COTS hardware is all that organizations need to run VNFs, so they help avoid vendor lock-in and proprietary hardware that is expensive to configure and deploy and can easily become obsolete. NFV allows standard hardware to run network functions, replacing dedicated hardware.
NFV is agile, shortening the time-to-market period by allowing for quick changes to the network infrastructure in support of new organizational goals and products. NFV networks also adjust more rapidly to fluctuations in traffic and demand. NFV networks scale the resources provided to them and the number of active VNFs up and down automatically using SDN software.
Challenges in network function virtualization are centered in three aspects of the approach: the VNFs, the NFV manager (NFVM), and the NFV infrastructure (NFVi). Because these three components are so closely interwoven, implementing NFV at scale is complex.
Complexity arises at three points of integration: when the NFV manager and existing computational infrastructure integrate, when the NFV manager and the VNFs integrate, and when various components of the NFV manager must coordinate their activities. To resolve the complexity and allow network function virtualization design elements to innovate freely, simplify these three points of integration.
The fact that multiple organizations have worked to standardize NFV tools over time has driven this complexity. This has evolved into a patchwork of approaches and standards—another of the disadvantages of network function virtualization.
Network Function Virtualization vs SDN
SDN and NFV are not dependent on each other, although they share some traits. Both use network abstraction and rely on virtualization, but they abstract resources and separate functions differently.
The difference between network function virtualization and SDN is that NFV refers to network component virtualization, while SDN refers to network architecture that decouples forwarding functions and network control, injecting programmability and automation into the network.
SDN and NFV together create a network that is built, managed, and operated by software. SDN separates network forwarding and control functions to achieve a centralized, programmable network control. NFV virtualizes network infrastructure, abstracting network functions from hardware. SDN software can run on NFV infrastructure, and together SDN and NFV can create a flexible, resource-efficient, and programmable network architecture.
To implement network function virtualization, create and deploy virtualized network functions, or VNFs. VNFs must be strategically built out in sequence as part of a service chain to deliver more complex products or services.
The orchestration process is another aspect of implementing NFV. The orchestration layer of a network must instantiate and monitor VNF instances, and repair and bill for them. These carrier-grade features enable scalable, highly reliable services, reduce maintenance and operational costs, and provide high security and availability.
It is critical that the properly implemented orchestration layer be able to manage VNFs without regard to the underlying technology. In other words, an orchestration layer must be able to manage any sort of VNF from any vendor running on any technology.
Reliable, high performance servers are a central piece of NFV equipment.
NFV architecture relies on server virtualization technology, and the virtualization layer options today are VMware, OpenStack, and container technology. VMware and OpenStack are the main hypervisor options. Container based network function virtualization, while not as widely deployed, offers next generation applications performance benefits.
MANO layer network function virtualization architectures vary, some open standard, some vendor supplied. The principal open source MANO option comes from the Linux Foundation, the Open Network Automation Platform (ONAP). Network operators must customize MANO to meet the specific requirements of their billing and operations architectures.
In the application layer, VNFs provide feature-rich network application code. In more sophisticated settings, network operators will select multiple VNFs from many to be service chained to deliver an expansive network function.
NFV is applicable across a broad scope of network functions, including mobile networks. Some common applications of network function virtualization include:
- Content delivery networks (CDN), including content delivery services, such as video streaming
- Evolved packet core (EPC)
- IP multi-media subsystem (IMS)
- Network monitoring
- Network slicing
- Load balancers
- Web Application Firewalls
- Security functions, including intrusion detection and prevention systems, firewalls, and NAT
- Session border control (SBC)
- Software-defined branch and SD-WAN
- Virtual customer premises equipment (vCPE)
Does VMware NSX Advanced Load Balancer Provide a Load Balancing Solution that Enables NFV?
VMware NSX Advanced Load Balancer makes turnkey, scalable load balancing for NFV deployments possible. Leave proprietary, purpose-built appliances behind with an application delivery controller/load balancer solution that allows NFV to live up to its promise. The VMware NSX Advanced Load Balancer architecture separates the control and data planes for application services such as load balancing and web application firewall and delivers load balancing as a flexible pool of resources that can run in any cloud environment.
With VMware NSX Advanced Load Balancer, a single REST API call provisions new load balancers within seconds and configures virtual services instantaneously. The platform also gathers and processes application data to present insightful security, performance, and end-user analytics about applications.
Learn more about how VMware NSX Advanced Load Balancer enables network function virtualization with the Cisco ACI Integration.
For more on the actual implementation of load balancing, security applications and web application firewalls check out our Application Delivery How-To Videos.