Network Functions Virtualization is Poised to Drive Innovation and Change in the Metro Service Edge

Apr 21, 2014

Communications service providers (CSPs) have begun actively evaluating network functions virtualization, or NFV, as a way to reduce costs and accelerate service innovation.  NFV represents a bold step towards software-defined services and programmable networks, and one that will impact all aspects of the communications industry for years to come.

NOTE: This article originally appeared under the title of “Network Functions Virtualization – it’s Here” in OSP Magazine.

NFV – What is It?
NFV was introduced to the industry in late 2012 by a group of thirteen CSPs who believed that today’s general purpose computing platforms – mainly Intel X.86-based servers – could be used to run many of the specialized functions that were currently implemented in dedicated hardware appliances.

These functions include firewalls, edge routers, test equipment, load balancers, border gateways, and others.  The idea is quite simple: separate the software from this proprietary hardware, and install the software to industry standard platforms.  The virtualized network functions (VNFs) will run on a modern dynamic cloud data center environment consisting of multi-core servers partitioned as virtual machines (VMs).

With nearly all traffic originating and terminating in the same metro area, the target for much of this innovation is focused on the metro service edge – the part of the network that spans from the metro core to the business site, cell tower or neighborhood edge.  So, the first implementations of NFV will be hosted in metro PoPs or in central offices where services are delivered over the Carrier Ethernet infrastructure and where the maximum benefits can be realized.

Why NFV?
By virtualizing these network functions, service providers will realize a number of benefits, including:

  • The ability to turn up and tear down a function dynamically.  This allows the CSP to use server resources only when the function is needed.  Once the resources are freed up, they can be used for another function.  With NFV, a test function could be turned up while the test is being performed and then turned down after the testing is complete.
  • The ability to pay for a function only when it is needed. Today, many appliance vendors offer their solutions in one package, intended to satisfy a broad range of performance needs.  This one-size-fits-all approach does not work well in all environments.  With NFV, the CSP pays only for the performance needed.
  • Taking advantage of the price-performance improvements that have followed the X.86 architecture over the years.  The cloud and data center industries have long enjoyed this benefit, and with NFV, the CSPs are able to do the same.
  • Reducing operational costs associated with sparing, maintaining and installing proprietary hardware for each individual network function.  With NFV, for example, CSPs can install multiple VNFs on the same set of general-purpose hardware and avoid the truck roll that would typically be required to install a new appliance.
  • The ability to quickly create new services through software only innovations.  With a standard hardware infrastructure in place, the CSP can leverage a community of independent software vendors, open source code, and home grown innovation to create revenue-generating services that were never before possible or practical.

NFV Use Case Example
In much of the industry discussions today, NFV is closely linked to software defined networking (SDN).  This is for good reason.  To achieve optimal results, the NFV implementation must take advantage of SDN concepts.

NFV’s carrier-class service orchestrator software uses SDN to automate the control of both physical network switching elements and virtual computing resources.  This orchestration software can receive a service order from a back office system and automatically create an optimized NFV-based service.

As an example, consider an enterprise service consisting of managed router and managed VPN components.  Before NFV, these services would be delivered using two or three dedicated appliances installed on the enterprise customer’s site and connected to the CSP’s network across a Carrier Ethernet connection (see Figure 1). If the Enterprise IT manager needed an additional service such as a managed firewall service, the CSP would dispatch a technician on-site to install, configure and test this new device.

Overture NFV - Managed Enterprise CPEWith NFV carrier-class service orchestrator, the enterprise IT manager simply logs onto a self-service portal, orders the additional managed firewall service, and the NFV software takes it from there.  The portal sends a service order to the NFV system, which installs the firewall virtual function on an available virtual machine in the central office or PoP (see Figure 2).  It then reconfigures the metro aggregation switch to steer the appropriate packet flows to the virtual function, and updates the billing record in the back office billing system.

Overture NFV - Virtual Managed Enterprise CPE Had there been any resource constraints or performance requirements, the NFV service orchestrator would have optimized for those constraints. In this simple example, the automation simplifies operational processes, speeds service delivery, and optimizes the available resources – saving the CSP time and money while providing a better end customer experience.

Both NFV and SDN bring interesting capabilities to the CSP services. However, they are most valuable when they are combined using scalable and on-demand cloud technologies.  Such combinations unleash service creation, activation and assurance for forward-looking service providers.

Realities of NFV in 2014
For all of its promise, NFV is still an immature technology.  Service providers are looking for open implementations yet many vendors are introducing proprietary closed systems.  This may stifle implementations.  In addition, NFV introduces a number of fundamental challenges to the CSP’s operational processes.  For all of these reasons, 2014 will see few large-scale commercial deployments of services based on NFV.   However, most major service providers are committed to smaller scale commercial trials.

The lessons learned in 2014 will form the foundation of a much more substantial roll out in 2015. OSP engineers field and technicians will benefit from learning more about virtualization, cloud and data center technologies now, as these professionals will begin to see changes to their workflow within the next two years. In the next couple years, there will likely be fewer truck rolls to customer sites, and in five years, there could be a dramatic reduction in installation requirements as many of these services will be installed by downloads.

To get a sense of where virtualization will likely initially take hold, consider the top use cases identified in the ETSI NFV Use Cases document (ETSI GS NFV 001 V1.1.1. 2013-10).  In this document, issued in October of 2013, the top targets for virtualization are the following:

  • Virtualized enterprise customer premise equipment
  • Virtualized of the provider edge router
  • Virtualized mobile core network and IMS
  • Virtualized mobile base station
  • Virtualized home environment
  • Virtualized content delivery networks
  • Virtualized fixed access network.

With the promise of driving cost out of and speed into the service delivery lifecycle, NFV has captured the imagination of service providers and their suppliers alike. This transformation to automation and virtualization will take several years to fully realize, but it will very likely impact nearly every aspect of the CSP business, from sales to the outside plant technician.

Network Functions Virtualization is Poised to Drive Innovation and Change in the Metro Service Edge was last modified: June 30th, 2015 by Overture Networks

Contact Overture to learn how you can increase profitability by transforming service delivery.