A network functions virtualization (NFV) cloud is a data center and network built to host, deploy, and service virtual network functions (VNFs) using a cloud network.
Prior to the advent of NFV, operators built application-specific networks using proprietary hardware. For example, to deploy a firewall service or a WAN service, the operator installed specific customer premises equipment (CPE) to deliver the service. Using an NFV software model, the operator can install NFV servers and software in the data center and then extend those VNFs and services to the customer using software.
Essentially, the advantage of using an NFV cloud is that service providers can more quickly roll out new services and VNFs using software rather than buying and installing specialized hardware.
NFV Cloud Standards and Servers
The basic model for building an NFV cloud is to take industry standard servers and build applications on top of them, which can then be accessed by customers using cloud software and Web provisioning. Several standards have been developed to specify the NFV architecture, including NFV MANO, a working group of the European Telecommunications Standards Institute Industry Specification Group (ETSI ISG). The ETSI-defined framework governs the management and orchestration of all resources in the cloud data center, including compute, networking, storage, and virtual machine resources. The ETSI MANO architecture was originally a white paper released at a conference in Darmstadt, Germany in 2012.
Although the ETSI MANO architecture has been very influential in how NFV clouds are built, many analysts and critics have pointed out that additional orchestration and management functions are needed. NFV requires orchestration software to set up VNF services, as well as management and operations software that can monitor them, repair problems, and bill for the services.
This has led to the call for another layer of software, known as Lifecycle Service Orchestration (LSO), that can carry out these important management and orchestration features and integrate with legacy Operations Support Systems (OSSes). MEF (the Metro Ethernet Forum), an industry association that counts over 200 companies as members, is leading the charge in LSO production and standardization. Members include AT&T, Cisco, Juniper Networks, Logitech, and ZTE. One thing is clear: The need for more flexible and agile VNFs will drive more and more network applications into the cloud over time.
Servers used for a cloud service. Source: Pixabay
VNF Has Limits and Advantages
The NFV cloud model is becoming increasingly more popular as network operators and service providers look for more agile and flexible ways to launch new services, especially enterprise services using VNFs.
Enterprise VNF services can include contextual routing, SD-WAN access, next-generation firewall, and multi-tenancy among others.
There are some criticisms about basic VNFs and their limitations, including:
- High consumption of hardware.
- A lack of universal standards, which limits where they can run.
- APIs typically do not include auto-scaling for changes in network demand.
- Lack of support for multi-tenancy.
Cloud-native VNFs resolve some of those issues, with the main distinction being self-management and scalability. New APIs are responsible for many updates in cloud VNFs, particularly with automation as there is now automated installation and configuration, scaling, monitoring, and updating.
Updated February 2019