NFV basics: A guide to NFV implementation, challenges and benefits
A comprehensive collection of articles, videos and more, hand-picked by our editors
Network Functions Virtualization (NFV)–the technology that allows operators to virtualize network functions and run them on less-expensive off-the-shelf servers–has generated quite the buzz with the promise of cost savings.
Yet despite initial dreams of tossing out expensive proprietary routers for commodity servers, NFV's path to network spending reduction is fuzzy. It remains unclear whether the technology will ever offer the performance necessary to replace proprietary routing and other core network functions in large-scale networks.
However, NFV will enable the virtualization of plenty of other network features, and will lead to a new level of service agility. With that, the technology promises an even more enticing promise than operational cost savings: It could mean new service revenue.
The idea is that NFV will let operators virtualize network appliances and services, which can then be dynamically provisioned and integrated into a larger orchestration context. Essentially, network services such as firewalling and load balancing will be provisioned as flexibly as the applications they support.
Until now, operators have been able to automate applications that live on virtual servers, but then they would have to manually provision the supporting network services, sometimes even on a hop-by-hop basis. This manual process has been counterproductive to automated, orchestrated application provisioning.
NFV is still in its early stages, but if the technology eventually enables dynamic provisioning of network services, it could become the dominant network technology shift of the decade.
Service agility, the holy grail of NFV
When it comes to NFV, operators love to toss about the term service agility. That's because currently it can take between two weeks and two months for operators to provision business services. Worse, if these services require new supporting infrastructure, it can take more than six months of lab work and then additional time to build the new devices. With NFV, operators hope to avoid this lengthy deployment because services are created through software integration and can be validated and deployed faster.
While NFV could clearly substitute for appliances like firewalls and load-balancers, or even for internal components of IP Multimedia Subsystem (IMS), the catch will be in how much of a provisioned service is actually software-based. The more legacy equipment involved in provisioning a new service, the less impact NFV can have on service agility. And that will be a common dilemma, since operators won't likely replace access equipment. They won't, for example, replace carrier Ethernet with hosted Ethernet switching. Even where an operator uses virtual switching, optical transport and other deeper OSI functions won't be converted into virtual functions. So in order to maximize NFV benefits, operators must extend NFV orchestration outside the boundaries of network functions to underlying legacy network. Vendors and operators are still working on strategies to extend NFV orchestration in this way.
New network features for potential revenue
While there are still challenges in using NFV to impact service velocity in a significant way, the technology can already enable a new range of service features that will be used to produce revenue. NFV will allow operators to offer context-aware network services, or the ability to ensure varying levels of service and performance according to application, user or location. Ultimately, providers will be able to charge accordingly for these granular services and performance assurances.
More on NFV
An NFV primer
Will NFV pave the road to scalable networks?
The ETSI specifications for NFV make it clear that NFV and cloud computing, or hosted applications, are very closely related. A network function could eventually be composed and provisioned in exactly the same way as a SaaS application. This could create a totally new model of services, blending network features with cloud computing or blending cloud-hosted application components with network services.
An example of this would be distributed load-balancing for cloudbursting. In a model in which copies of an application are hosted in the cloud, but are dispersed and shared among users, NFV and its dynamic provisioning can be used to load balance effectively across sites and users.
This kind of granularity in management lets operators define QoS/QoE goals for a cloud application according to users' specific needs and meet those goals at the network level. Additionally, they can integrate managed security and availability features into applications dynamically. Network operators will use this capability to differentiate their own higher-level services by tying them directly to network behavior.
NFV services must offer this kind of integration of applications and network behavior in order to be differentiated from other cloud services. But there are also other important features beyond transport/connection QoS.
Network access security, as well as network resiliency and manageability, are all features that can be linked to the network and offered more easily and granularly with NFV. NFV's dynamic distributed load-balancing and management across data centers or clouds will allow operators to secure sites from DDoS attacks, as well as to manage an array of remote devices, and even to use IT tools for facilities management, to make power usage decisions for example.
NFV could also be extended to non-connectivity features of the network. Probably the most relevant of these use cases is context-aware networking or the ability to make performance and provisioning decisions based on location, user activity and even social context. If users are busy, they don't have to receive application access. Or they can receive varying levels of access depending on the type of application they need.
Ultimately, it's not likely that NFV will introduce new services that couldn't otherwise be provided. NFV is about function hosting, and there are many ways to host functions—from appliances to dedicated servers, through virtualization and via the cloud. But NFV will create a framework for deploying complex applications and operating them with high reliability and low operational expenditure costs. This will allow operators to price even complex contextual applications at levels that are consistent with broad adoption and still make a satisfactory profit from their investment. It will also allow operators to bring new services to market faster, accelerating revenue realization and encouraging investment.
The future of carrier networking depends on a combination of utility for buyers and profit for the operators. NFV doesn't change the game, but it optimizes how it's played, and that's just as important in today's world.