Open Source Software and SDN

Light Reading held its third Big Communications Event (BCE, formerly BTE) in Austin last week. Packet Design was a demo sponsor for the two-day event following our appearance at the celebration dinner after being named a finalist for the Most Innovative SDN Product Strategy Leading Lights award. I was able to attend some of the keynote sessions, which featured service provider industry luminaries, as well as a few interesting breakout sessions. Rather than provide a review of everything I learned, I thought I would focus on one topic: the role of open source software as an SDN enabler for network operators.

As we have documented many times in this blog, Packet Design is both a participant in, and supporter of open source initiatives. Notably, our SDN Management and Orchestration solution leverages open source SDN controllers to enable automated network configurations and intelligent provisioning of services. We are an active source code contributor to the OpenDaylight (ODL) project, and Neela Jacques, its Executive Director, keynoted at our Customer Symposium in Paris earlier this year. So I was interested to hear what others had to say on the topic. I had the opportunity during one breakout session that focused on the value of open source to network operators.

The session was a panel discussion moderated by Senior Heavy Reading Analyst, Roz Roseboro. The panelists, who included representatives from the service provider industry, vendors, and the open source community, were: Mark Collier, COO of the OpenStack Foundation; Lauren Cooney, Senior Director of Strategic Programs in Cisco’s Chief Technology & Architecture Office; Heather Kirksey, Director for the Open NFV (OPNFV) project; Randy Nicklas, EVP of Engineering & CTO at service provider Windstream; and Dan Pitt, Executive Director of the Open Networking Foundation. Here are my key takeaways from the lively discussion.

Why open source vs. proprietary solutions?

While lower cost products may seem like the obvious benefit for network operators, Nicklas from Windstream said open source economics are not yet proven, although he acknowledged that they should help to increase competition and keep prices low. However, everyone agreed that the pace of innovation in a software-defined world is accelerated, best practices shared more easily, and duplication of effort minimized when open source projects deliver solutions that would otherwise be provided commercially. Vendors can bring products to market faster by leveraging open source for capabilities that do not require differentiation. To underscore this point, Cooney stated that Cisco is currently involved in more than twenty open source projects and wants to integrate open source with proprietary technology across all of its solutions.

It is easy to think of open source projects in the same vein as standards bodies, but if we consider the work being done in the leading SDN and NFV open source projects, the watchword is rapid innovation and iteration. Getting new capabilities into the hands of vendors and consumers so that they can be tested, deployed, and delivering business value quickly trumps the inherently time-consuming standards-definition process. As a consumer of ODL outputs, Packet Design has seen multiple, substantial software releases in the last couple of years. In effect, because of their broad industry involvement, these open source projects are creating de facto standards.

Why and how should network operators engage with open source communities?

Despite the benefits, with the exception of a few larger operators, service providers and enterprises are poorly represented in the SDN and NFV open source projects. Given the network operating costs and capital expenditures as a percentage of overall company budget, operators should be more involved and demand that their vendors leverage open source where possible. Nicklas noted that vendors are not always as willing to work together as customers like Windstream would like. I agree, but in my opinion, the current momentum of leading SDN/NFV open source projects provides a compelling reason for vendors to participate, or risk being marginalized. There will always be opportunities for vendors to deliver proprietary value-add capabilities.

ONF’s Pitt noted that there are 2.5 million open source projects and most are failures – typically due to poor community organization and support. Cooney advised evaluating organizational maturity and to avoid those open source projects that seem to have a lot of “drama.” OpenStack’s Collier agreed that the number of open source options can be confusing. He suggested taking the lead from others with similar needs, and working through a vendor partner involved in the open source project to get to know how things work. Kirksey added that open source groups like OPNFV publish reference architectures, matrices of known working configurations, and other tools to facilitate user adoption. Even so, Pitt said that network operators need to act more like software companies in the way they develop solutions and work with vendors. Cooney recommended looking online for educational content and not feeling compelled to attend the myriad open source conferences.

What is the ultimate goal for open source projects?

To wrap up the session, Roseboro asked the panel how they would measure the success of open source projects. Collier said that his goal for OpenStack is to achieve 50 billion devices connected via an open source enabled network. A lofty goal indeed.

Nicklas, offering the service provider’s perspective, stated that success needs to be measured in dollar terms—improved efficiency, capex savings, higher margins on services, faster service time to market, and increased responsiveness to business and customer needs.

Pitt was perhaps the most succinct. Success for network operators will be the ability to offer better services to their customers, he said; to solve a problem so they can move on.

Amen to that.