1. Virtual Network Function (VNF): This is a fundamental part of the NFV architecture. It simply denotes a virtualized network element. An example is when a router is virtualized it is called Router VNF.
2. Element Management (EM): This is a system that caters for the functional management of VNF i.e. Fault, configuration, Accounting, performance and security management.
3. VNF Manager: This is responsible for managing a single VNF or multiple VNFs. It is responsible for setting up and shutting down VNFs
4. Network Function Virtualization Infrastructure: This is the environment in which VNFs are run. It also consists of physical resources, virtualization layer and virtual resources.
4.1 Compute, Memory and Networking Resources: This is the physical aspect of the NFVI where virtual resources are instantiated on physical resources.
4.2 Virtual Compute, Virtual Memory and Virtual Networking resources: This is the virtual aspect where abstraction of physical resources into virtual resources occurs and is used by VNFs
4.3 Virtualization Layer: Decoupling of software from hardware occurs at this layer commonly referred to a “Hypervisor” in the industry.
5. Virtualized Infrastructure Manager (VIM): This is responsible for the management and control of the compute, network and storage of the NFVI resources within one operator’s infrastructure domain.
6. NFV Orchestrator: This is responsible for generating, maintaining and tearing down of network services of VNF
7. Operation Support System/ Business Support System (OSS/BSS): OSS deals with network management, fault management and service management. BSS deals with customer management, product management and order management.
NFV Uses Cases
NFV Versus SDN Versus Traditional Networking
In the old days, the earliest means of data transport networks were traditional phone networks and telegrams. To design such networks, design requirements included latency, throughput, availability and the capacity to carry data with as little loss as possible. These factors impacted the development of hardware and equipment required to transport data. Also, specific use cases were used to build the hardware devices whereas making them only for specific uses. With the arrival of data transport networks, there are no changes to the factors that influence network design or devices.
All traditional network devices were created for a specific purpose and the data networks constructed were designed to meet the efficiency criteria effectively, The software running on these devices are tightly coupled to it and have FPGA or IC’s integrated into them which keep them dedicated to performing their specified function.
As the demand for bandwidth increases in the enterprise which is heavily driven by IoT application, videos and mobile, Internet Service Providers continue to find alternatives on how to expand and scale up their network services, most likely without increase in costs. The qualities of traditional devices present a blockage to this requirement and present many limits that discourage scalability, deployment costs, and operational efficiency. Some of these limitations include flexibility limitations, scalability constraints, manageability issues, high operational costs, interoperability and migration considerations.
With Datacenters, server virtualization has become proven technology where virtual servers running on shared hardware have replaced stacks of specialized independent server hardware. This is the concept that NFV builds on “Server Virtualization”. It stretches the concept beyond just servers and also touches on network devices. It encourages managing, monitoring, provisioning and deployment of virtualized network entities.
NFV has been used to reference and ecosystem that consists of virtual network devices, management tools and infrastructure that allows for integration of these pieces of software with hardware. A more accurate definition for NFV would be “a method and technology that replaces physical network devices performing specified functions while running on generic computer hardware”. Virtualization has developed to the stage where physical devices can be masked and COTS (commodity of the shelf) boxes can be used to provide infrastructure for NFV. As regards traditional network, vendors are not concerned with hardware on which the code will run, due to the fact that the hardware work as a dedicated equipment developed, customized and deployed for the specified network function. Complete control exists over both hardware and software running on the device that allows for flexibility to design the hardware and its performance requirements based on the part the device will paly on the network. In the case of virtualized network functions, unrealistic assumptions cannot be made about the ability the hardware has to offer. With appreciation to SDN, NFV involves decoupling of software to hardware and proves to show the ability to use commercially available hardware to execute the virtual offerings of very specific network functions.
NFV in the Telecommunications Industry
NFV Related Concepts
Cloud Computing: Within the cloud environment, the role of the service provider is bi-faceted. The infrastructure providers who handle the management of cloud platforms and lease resources based on a pay-as –you-use model and the service providers that rent resources form infrastructure providers to aid end users. Five important characteristics and three cloud models make up the cloud model.
Characteristics
a. On-demand Self-Service: This involves a consumer single handedly provisioning computer capabilities, which vary from network storage to server time as needed automatically without any human interaction with service provider.
b. Broad network access: This involves accessing the capabilities over the network whereby access is gained through standard mechanisms that promote various thin or think clients.
c. Resource Pooling: Using a multi tenant approach, providers’ resources are pooled to serve multiple consumers with varying physical and virtual resources assigned dynamically to meet consumers demand.
d. Rapid Elasticity: Based on demand, capabilities can be scaled up or down which in most cases is done automatically.
e. Measured Service: A metering capability is employed by cloud systems to automatically control and optimize resources.
Models
a. Software-as-a-service: Here, users are able to use applications made available by provider running on a cloud infrastructure. These applications are accessible via a thin client interface or client devices
b. Platform-as-a service: In this model, user is able to deploy to the cloud infrastructure consumer-created or acquired applications created using programming languages.
c. Infrastructure-as-a-service: The user is able to provision fundamental computing resources such as storage, networks and others where consumer is allowed to run any choice of software which may include operating systems etc.
Relationship Between Cloud computing and NFV
Generally, NFV is not limited to functions for services in telecommunication. Many applications already run on COTS devices. However, since the most promising use cases for NFV originate from the telecommunications industry, and because performance and reliability requirements of carrier-grade functions are higher than those of IT applications, NFV performance should be carrier-class. The flexibility that cloud computing offers makes it the best option that presents an opportunity of achieving efficiency and cost reduction that are motivating Telecommunication Service providers to move towards NFV.
Moreover the deployment of network functions in cloud will likely effect a change in every aspect of how services are developed and deployed. While work continues to be done with reference to networked clouds and inter-cloud networking, telecom networks are different from cloud computing networks in these 3 ways:
i. Workloads of data plane in telecommunications network mean high pressure on performance
ii. Network topologies of telecom networks place harsh demands on the network and a need for global network view exists for management
iii. The telecom industry requires reliability, availability and scalability.
In the telecom networks, the site infrastructure makes this set of features available.
If NFV is to be based on cloud computing, a replication of these features has to be done by the cloud infrastructure in a manner that they can be orchestrated.
Research on Cloud-Based NFV
Within cloud computing environments for NFV to perform acceptably, the underlying infrastructure needs to make available some certain functions, which involve orchestration, scheduling, networking and monitoring capacities. Openstack that has been identified as one of the major components of a cloud based NFV framework does not meet some key requirements. One of such requirements includes QoS requirements. OpenANFV proposes a framework based on Open stack that uses hardware acceleration to enhance the performance of virtual network functions.
Software Defined Networks
This is an area getting great attention from both industry and academia because of the architecture it recommends for the management of large-scale complex networks. SDN is responsible for decoupling of control plane from forwarding functions which allows for the programmability of network controls via arbitrary open interfaces and also making the underlying infrastructure packet forwarding devices that are programmable.
Relationship between SDN and NFV
NFV and SDN have a lot in common as they both promote a change towards open software and standard network hardware. Most importantly in the same way that NFV aims at running network functions on standard hardware, SDN’s control plane can be implemented as software running on standard hardware. Both SDN and NFV aim to take advantage of virtualization and automation to achieve their respective goals.