For this week’s interview, we turn to Matt Davy, Director of InCNTRE at Indiana University. We had a chance to speak with him when he was out here for the ESCC/Internet2 Joint Techs meeting. He shared with us the keys to building InCNTRE into a well-respected leader in the whole SDN movement and describes his plans and vision for SDN at InCNTRE.
SDxCentral: What’s InCNTRE?
Matt: The Indiana Center for Network Translational Research and Education, or InCNTRE, is a center at Indiana University with a mission to advance development, increase knowledge, and encourage adoption of OpenFlow and other standards-based SDN technologies. To achieve this mission, we’ve established three main areas of focus: research, education and testing.
SDxCentral: Both you and the InCNTRE team have been heavily involved in OpenFlow from the early days, how did you get started?
Matt: I recently did a search of my mail archives to see when we first got started with OpenFlow. Our work on OpenFlow dates all the way back to March of 2008, shortly after the first public demo and talk on OpenFlow. At the time, Stanford University was looking to move OpenFlow beyond their own lab and deploy OpenFlow testbeds on several university campuses. The National Science Foundation provided funding to Stanford and 7 other campuses, including Indiana University, to deploy OpenFlow testbeds as part of the GENI project. When we started deploying OpenFlow in 2009 we immediately realized it’s potential value. Indiana University immediately began investing it’s own resources and quickly established a clear leadership role in this area. We now have over 25 people working on SDN projects.
SDxCentral: 25 people, wow! That’s impressive. So what SDN projects are active at InCNTRE?
Matt: We’re probably best known for our SDN Lab which is the only test lab focused solely on testing SDN implementations. We also have a summer internship program called the Summer of Networking that provides ten weeks of intensive hands-on networking education to undergraduate and graduate students with a particular focus on SDN. Last, but certainly not least, we have a number of interesting network research projects related to SDN led by our director Dr. Martin Swany.
SDxCentral: What role do you think InCNTRE plays in the SDN community?
Matt: I’ve heard numerous people in the networking industry describe us as thought leaders in the SDN community. We strongly support the ONF’s vision to make SDN the new norm for networking and are helping the community achieve that vision in any way we can. One way InCNTRE is contributing is by providing resources to the community that are critical to SDN’s success such as testing and education services, often long before the commercial market is ready to provide these services. We’re also one of the first organizations developing and deploying OpenFlow solutions in production networks in partnership with Indiana University’s GlobalNOC organization. IU is partnering with Internet2 to deploy a national 100G production SDN backbone later this year. So we’re sharing this valuable early experience with the community to help shape the future of the technology.
SDxCentral: Aside from Internet2 activities and your active projects, what other services does InCNTRE provide?
Matt: The first service we established well over a year ago was an SDN interoperability testing consortium. Similar to interop testing consortiums for other networking technologies, companies can join the consortium and have InCNTRE test their SDN implementations for interoperability with other implementations. More recently we started providing hands-on training for OpenFlow. In the first two months of the program alone, we’ve already provided high-quality, hands-on OpenFlow training to more than 100 network engineers from over 50 different companies and organizations. We also provide custom SDN testing services that can be used for proof-of-concept testing by companies interested in deploying SDN solutions or networking vendors that are not part of the interop testing consortium. More recently, we have been consulting with companies that are trying to understand the SDN technologies, the use-cases and how it might fit into their future network architecture.
Also, we have been very active in the Open Networking Foundation since it’s creation and are applying to become a certified testing facility with the ONF’s. So we plan to start offering OpenFlow conformance testing services later this year.
SDxCentral: Why and how would companies participate at InCNTRE? What value do they get out of it?
Matt: Companies can participate in our interop testing consortium simply by signing up for a membership and paying an annual membership fee. Consortium members share in the lab’s valuable resources that currently includes 4 full-time experienced test engineers, several graduate assistants, network testing equipment, and OpenFlow controllers and switches from several vendors. Members also have access to the suites of test plans we have developed that verify interoperability of all components of SDN solutions including switches, controllers and applications. Ultimately this gives our members the ability to demonstrate to their customers that their OpenFlow implementation works as part of a complete solution with implementations from many other companies. And most importantly this is validated by a neutral 3rd party lab. Of course for all the services I mentioned, you can contact us at firstname.lastname@example.org to get more information or visit our website at incntre.iu.edu.
SDxCentral: Sounds compelling, who do you have so far that have signed up?
Matt: Our consortium members include Big Switch Networks, Brocade, Cisco, HP, IBM and NEC. We have also received significant sponsorship support from Opengear, Panduit, HP, Matrix Integration and Vello Systems. The testing consortium is growing rapidly and expect to announce four or five new members in the next few weeks as well as several more members by the time we host the ONF’s interop event in our lab in October.
SDxCentral: Impressive growth, but Indiana isn’t the first place one would think of as a networking powerhouse. How have you been able to recruit the talent needed to offer those services?
Matt: Indiana University is perhaps the best kept secret in the networking industry. In 1998, Indiana University created an organization called the GlobalNOC to manage a new national network being deployed by Internet2. 14 years later, the IU GlobalNOC now employs nearly 100 staff and provides a full suite of network management services to something like 20 research and education networks throughout the US. Because of our unique role in designing and operating cutting edge research networks in a laid-back university environment, we’ve been able to recruit some of the top network engineers and network software developers in the industry. We’ve also developed deep relationships through the networking industry. We’ve leveraged both our internal resources and our relationships to recruit top networking talent into InCNTRE.
SDxCentral: Why isn’t UNH doing it? Isn’t this a natural extension of their interoperability labs?
Matt: It’s important to note that OpenFlow is not a network protocol like OSPF or BGP or a data-plane technology like 10 GbE or WiFi. OpenFlow is a software interface between the components of a distributed software system. When we first started testing OpenFlow implementations nearly 3 years ago, it was clear we needed people with expertise in software testing for distributed systems as well as people who understood networking and network protocol testing. Also, our members have told us that a key to our value proposition is that, in addition to testing, we’re actively developing open-source OpenFlow software and deploying it in real production networks. So we can share this experience with our members, apply it when building test plans, and help move the technology forward. I’d also note that, in researching the history of several well-known test labs, I discovered that in nearly every case the labs started at a university that was an early adopter of a new networking technology.
SDxCentral: What about commercial test labs? Are they getting into the act too?
Matt: I’ve heard that there are commercial labs that are considering starting testing programs for OpenFlow. I think our value proposition is very different than a commercial test lab. Our mission is to advance SDN through multiple initiatives. Most of the vendors that have joined our SDN testing consortium have also engaged with us on professional OpenFlow training, collaborative research, supporting and ultimately recruiting our grad students, as well as other SDN initiatives.
SDxCentral: What are your goals for these services and InCNTRE over the next few years?
Matt: We see the demand for testing and training services growing significantly over the next couple years, so we’re definitely gearing up to expand in those areas. However, we’re also very focused on more future looking problems, perhaps further out then industry is focused on. We believe that, if SDN does in fact become the new norm in networks, we will need a new breed of network engineer/developer – one whose just as comfortable writing Java or Python code as today’s network engineers are configuring IOS or JunOS. To do this, I think we need more CS students graduating with deep, practical experience with networks. So we’re very focused on reworking the CS networking curriculum and creating internships to train the next generation of network engineers and developers. We’re also expanding our network research focus to use SDN to address the major challenges in networking over the next 3, 5 or 10 years.
SDxCentral: Do you think OpenFlow and SDN are a flash-in-the-pan? Will Cisco and OnePK eventually make it irrelevant?
Matt: In addition to leading InCNTRE, I also retain overall responsibility for network architecture for Indiana University’s enterprise network which includes over 120,000 users. For the past three years, each time we’ve run into a new challenge we needed to solve, we’ve looked at the existing products and technologies that were available, but also considered how we might solve the problem within the SDN/OpenFlow architecture. In every case, the solution was much simpler and more elegant when using SDN then without it. This leads me to believe there’s really something to this and it’s not going away any time soon.
SDxCentral: If you could make a request of the SDN community out there, what would it be?
Matt: I think Martin Casado said it best at the first Open Networking Summit last fall. SDN is going to succeed because of the community of people who are pitching in and contributing their time and talent to make it a success. It’s a technology that is being pulled forward by the customers, not pushed forward by the vendors. At Indiana University, we’ve been working hard to pull this technology forward for more than 2 years now. If you or your company buys networking gear, get up to speed on what SDN and OpenFlow are and get involved. Join the ONF, engage with InCNTRE, contribute to open-source projects, build a testbed!