Q&A with Heather Kirksey, Director of the OPNFV

Networking is proving to be one of the fastest-growing industries for open source software development, and a key element of that is Network Functions Virtualisation (NFV). Set to transform the way networks are designed, NFV — an initiative to move to the cloud network services that have traditionally been delivered by proprietary, dedicated hardware — has seen tremendous growth over the past few years, with upward trends expected to continue.

We spoke to Heather Kirksey, director of the Open Platform for Network Functions Virtualization (OPNFV) project, about why NFV and SDN are on the precipice of reaching critical mass and how and why this network transformation began.

The macro trends in network include both an explosion of data traffic and growing diversity in the form that traffic takes. In addition, the predictability of planning for different types of network traffic has dramatically decreased as new services spring up; for example, who could have guessed at the huge increase in unicast streaming traffic Netflix would cause before its service became so popular?

To address these issues, service providers are looking to make their networks more agile and more flexible. They are moving away from architectures that depend on an ever-growing number of single-purpose, vertically integrated network elements to architectures that make use of software-based innovations, such as Software Defined Networking (SDN), compute and storage virtualisation, and cloud-style elasticity and resource management.

Can you talk a little bit more about NFV, and how that relates to SDN and cloud?

Network Function Virtualisation (NFV) is the transformation from a network composed of proprietary hardware network elements to a dynamic, software-based network. Functions that once resided in hardware (for example, firewall or Deep Packet Inspection (DPI) appliances, Customer Premises Equipment (CPE), much of the traffic and subscriber management capabilities of a Fixed/Mobile network) can now be deployed as software applications on white box hardware. These applications are often referred to as Virtual Network Functions (or VNFs) and can be deployed and managed using typical software management capabilities rather than proprietary Element and Network Management Systems (EMS/NMS).

Both SDN and cloud are important foundational technologies that enable NFV. Cloud technologies such as OpenStack provide the hardware, VM or container, and resource management capabilities for the infrastructure, and SDN provides the network instantiation infrastructure (overlays, underlays, VPNs, etc.) and the network applications that can run on top of them.

Who are the early adopters of these new technologies and why do you see these groups as the pioneers?

A number of the large global service providers are looking to transform their networks to take advantage of these new developments. Some of OPNFV’s most active Service Provider members are AT&T, China Mobile, Orange, and NTT DOCOMO. They may differ in the specific applications or use cases they’re implementing first, but they have all indicated their commitment to the industry to move in this direction.

Haven’t telcos traditionally been slower to adapt to change than other sectors of the tech industry --what’s the impetus for change now?

The impetus is similar to what usually drives large-scale change: the status quo has become too painful to maintain. The telcos are facing a competitive environment unlike ever before with companies like Facebook, Netflix and Google outpacing them on the service innovation front while increasing the demands on their networks.

Meanwhile, several generations of advancing network capabilities (such as the stepwise progression to 4G) and introduction over time of purpose-built appliances to solve specific problems (such as DPI and CDN appliances) have left operators with complex, brittle, unmanageable networks. In order to remain competitive they are having to fundamentally rethink how they architect and deploy networks, with NFV becoming the key visionary concept to effect that change.

Why does any of this matter beyond the telco space?

Service providers are not the only organisations dealing with network complexity. Although their issues are particularly pressing, many enterprises are finding the overhead of managing large complex networks becoming untenable. Whether it’s managing all the Business CPE connecting branches to more central locations, growing real-time needs of Financial Services organisations, or the human capital cost of managing ever-growing DPI, firewall, load balancing, VPN, IDS, and other networking and security-oriented appliances, enterprises, web-scale companies and financial services organisations can also benefit from moving to a more software-oriented, agile network infrastructure as well.

It’s tempting to view open source only through the lens of 'no cost' software, but it’s best viewed through the lens of large-scale, industry level R&D. Every company developing solutions and solving the same base problems again and again in their own silos does nothing to advance the state of the industry and makes interoperability and common understandings more difficult to achieve. Open source, however, enables an environment where colleagues can collaborate together across company lines, solve the hard foundational problems, and free up their smartest talent to focus on value-add, new service creation, and even unexpected innovations.

What is OPNFV and why was it established?

OPNFV is a carrier-grade, integrated, open source platform to accelerate the introduction of new NFV products and services. That’s a bit of a mouthful, but essentially we are an open source organisation focused on enabling NFV for our members through open source collaboration. We were established because of a belief on the part of our founding members that in order to enable multi-vendor, interoperable NFV implementations in the timeframe demanded by the market, we needed an independently governed open source foundation that would enable us to focus on integration of existing upstream components.

This includes deploying and testing those components on a wide variety of infrastructures, openly discussing -- across company boundaries -- what additional features and capabilities were necessary to realise OPNFV, and then be able to approach other upstream communities (such as OpenStack, Linux, OVS, KVM, OpenDaylight and others) with one collective voice and common use cases.

What impact do you expect OPNFV will have on the industry all-up?

We expect OPNFV will help the industry make the journey to NFV transformation more quickly and perhaps with a bit less pain. We expect it to help make the new VNF applications more interoperable with many flavors of infrastructure. We expect to help upstream communities understand NFV use cases and help them make their technologies more suited for NFV. And ultimately, I hope that we can bring a more collaborative and problem-solving culture to the larger telecommunications ecosystem.

Tell us a little about your Summit taking place this week in Berlin.

This is our second-annual OPNFV Summit, a platform for the industry to come together to collaborate, innovate and explore NFV. The Summit attracts developers, end users and upstream communities who all work together with the common goal of advancing open source NFV. The week will feature keynotes, technical sessions and hands-on design workshops as well as impromptu un-conference style community gatherings, which fits well with this focus of this year’s Summit, 'Community and Collaboration'.

We’re also hosting a Design Summit (focused on developers), a technology showcase, PoC zone, Project Theater, and networking events, including an OPNFV Orientation to help newcomers to the project -- both technical and non-technical -- learn how to get started and become familiar with OPNFV and community. There’s not a better place to meet colleagues in the NFV space and work together with the community to make a difference in the future of networking. More details on the Summit can be found here.