NFV: A New Direction for OpenStack

Neela Jacques, Executive Director, OpenDaylight Project
219
339
67

Neela Jacques, Executive Director, OpenDaylight Project

People for years have been asking a fundamental question about OpenStack: Is it about pets or cattle? Or, put another way -- is OpenStack an open source VMware vSphere, or a way to create an AWS-like infrastructure?
 
  Carriers are a perfect early adopter for large-scale production deployments of cloud and Software-Defined Networking (SDN) 
 

The distinction is an important one. The feature set and functionality required to host a set of existing enterprise workloads is fundamentally different from that of cloud native workloads (“cloud native” refers to applications rewritten from scratch for a cloud infrastructure).

This has never been an easy choice to make. Today’s market opportunity clearly lies with hosting the world’s existing applications, but that’s like skating to where the puck is rather than where it’s going--in our case, the more modular and agile world of cloud. With a plethora of possible workload types and use cases, the OpenStack community has struggled to find a focal point or dominant workload to focus on.

Network Functions Virtualization (NFV), fueled by integration with other open source projects like OPNFV, provides OpenStack with a very clear set of needs, a large market with highly motivated customers able and willing to spend money.

Carriers Come Out in Force at OpenStack Austin

At the OpenStack Summit in Austin last month, carriers came out in force – AT&T and Verizon both delivered powerful keynotes, Canonical spoke about carrier-ready infrastructure, and--for the first time--the event featured a Telco and NFV track, and it was packed to the gills!

Carriers and service providers were highly visible, and nearly every session seemed to touch on NFV. In his opening keynote address, OpenStack Foundation Executive Director Jonathan Bryce declared telecommunication companies powering NFV with open source cloud software has been among the fastest growing use cases over the past year.

Why has the community jumped on NFV so fast? Because carriers provide the exact mix of attributes that can take OpenStack to the next level. One challenge with broad platforms is that there is a chicken and egg problem – users want to adopt feature-rich, mature, well-tested code, but those same users have to be active early and in sufficient quantities to provide the motivation, feedback, and real-world experience to get us there.

According to Infonetics Research, carriers collectively spend over $350Bn a year on CAPEX (to drive revenues of $1.997 Tn) – and they have a strong and pressing need for greater agility. With the rate at which data is growing and with the shift to on-demand applications and mobile usage, a transition to a more agile compute and network infrastructure isn’t a “nice to have” – it’s a necessity.

Thus, carriers are a perfect early adopter for large-scale production deployments of cloud and Software-Defined Networking (SDN). They have the need and resources to really invest in working with the open source community to refine and improve the technology to meet the needs of demanding carrier environments. AT&T’s approach to their network of the future work is a great example of this in action. At a smaller scale, companies like China Mobile and Comcast are leveraging open source platforms like OpenStack and OpenDaylight to bring to life cloud and NFV use cases.

NFV also provides a great focal point for the industry: a set of clear use cases with well-defined needs that can help the community converge on implementing requirements. Many VNFs will clearly start as simply virtual versions of their pizza box predecessors, yet they will be rewritten and redesigned over time to better take advantage of the virtualized environments.

NFV is also helping to resolve the debate regarding the need for an SDN controller with OpenStack. During the Summit, carrier after carrier articulated that at the scale they are deploying, the network agility an SDN controller provides is not optional.

Moderating a Panel of community devs and users from AT&T, China Mobile, Ericsson and Brocade during the OpenStack Summit 2016 in Austin: “Among the Cloud: Open Source NFV + SDN Deployments”

Carriers Are Investing Heavily in Open Source

End-user engagement in Open Source is critical. Without formal roadmaps, even with Product Managers with their Product Requirements Documents (PRDs) or even command-and-control leadership imposing its vision, how do communities of thousands of developers know which are the most important areas of focus? It is the presence of end-users at design summits, HackFests and even within projects that provides guidance and feedback.

We are already seeing this. Companies like AT&T, Comcast, and Verizon have built large OpenStack teams that are getting more deeply involved in the core developer community. The Open Platform for Network Functions Virtualization (OPNFV) project has provided a great forum for carriers to come together with the vendor community to test not only one component, but also the full set of components needed for NFV – including KVM, OVS, and the OpenDaylight controller.

Similarly, carriers are increasingly active with the OpenDaylight community; ten members of the OpenDaylight Advisory Group represent major global carriers. They are contributing heavily to projects such as the OpenDaylight performance white paper, and have now submitted four technical projects in the last two releases. (Learn more about OpenDaylight as an NFV Controller and Intelligent Application Sharing through Coexistence.)

The Trickle-down Economics of Cloud

A question often asked at this Summit was “What about Enterprises?” While not all of the needs of carriers will translate to those of enterprises, many will. Scalability, security, stability, and performance are universal needs. Integration between OpenStack and SDN controllers such as OpenDaylight are needed by anyone with high scale and the desire to manage both an OpenStack network and traditional networks with one common platform. This is highly attractive to many large enterprises as well.

More and more enterprises are also starting to look and feel more like service providers. Google, Facebook, and Amazon had to embrace cloud architectures and SDN before open source technology was even an option. The next set of web-scale giants is voraciously picking up open source.

A great example of this is in China with Baidu, Tencent, and Alibaba (collectively known as “BAT”) building everything they do on open source (see Tencent and Why Open Source is about to explode in China on Linux.com). Companies like Fidelity, Thomson Reuters, and Walmart.com all starting to walk down similar paths that we’ve seen with the major carriers. While smaller enterprises may simply outsource large swaths of their IT to service providers, many large enterprises’ compute, and network needs are becoming ever more similar to those of carriers. I believe much of the work being done by the carriers in OpenStack, OPNFV, and OpenDaylight will have broad benefits across the industry.

Read Also

The Mainframe isn't an Isolated Bunker in the Application Economy

Ashok Reddy,General Manager of Mainframe,CA Technologies

Efficient Ways to Manage the Exponential Growth of Data

Stephane Bourles, CIO, Brain Balance Achievement Centers

How No-Code/Low-Code App Platforms Bring Value to Your Business

Thierry Ciot, Software Architect and Product Owner, Progress

Make Cybersecurity Resiliency a Top Priority at Your Organization

Audie Hittle, CTO Federal and Cyber Security, Dell EMC Emerging Technologies Division