In a manner of speaking, it was tragedy that ultimately brought Robert Cannistra, senior professional lecturer of computer science and information technology at Marist College, to the world of academia and software-defined networking. As a technical services manager at vLink Solutions (formerly All Tech Data Systems Inc.) back in 2001, Cannistra was scheduled to be in the basement of the World Trade Center during the morning of 9/11. However, he was called down to the company headquarters, eight miles from the Pentagon, the night before.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
"Needless to say, it wasn't my fate at that time," Cannistra said. Prior to his time at vLink Solutions, Cannistra worked as a solutions architect at AT&T, designing, implementing and administering the company's Intranet. After leaving AT&T, he went on to help create the infrastructure behind About.com, where he worked before making the move to vLink.
Flash forward to a few months after 9/11, and Cannistra knew his time at vLink was coming to an end. "I had a change of heart," he said. "I continued with that company for a couple months, but the work I was doing in New York hit close to home. We were implementing the coding communications ring around the country at the time; some of the places we had to go were four levels deep under Manhattan. I had an awakening [after 9/11] and wanted to stay closer to home."
Cannistra was teaching part-time for the Cisco Networking Academy during late 2001 when he spotted an open position at Marist in Poughkeepsie, N.Y. The job was essentially helping to lead a lab and add some practical networking courses into the curriculum. "So I applied for the job and the rest was history," he said. "I still do consulting in addition to that, which helps to blend that practicality with theory in academia."
Discovering software-defined networking at Marist
Cannistra had followed research on programmable networks for quite some time before first hearing about the concept of SDN. It was a faculty grant awarded to Marist from IBM that jump-started Cannistra's work with the rising technology -- but not before he had students figure out what SDN was all about.
Do I think the end game is OpenFlow? No … but the concept of this protocol being open, accessible and providing intelligence from a central controller is pretty powerful.
senior lecturer, Marist College
"About two and a half years ago, IBM put together a faculty grant for us to do work with SDN and OpenFlow in particular, because the standards body was about to be developed, and this protocol seemed like a good idea," he said. Cannistra hired a student after receiving the grant and reading white papers on SDN and OpenFlow. "I said [to him], 'Take this and go look at it and see what's out there,'" he said. "He reported back a week later and said, 'This is so cool.'" Cannistra began diving into SDN himself. "I thought it seemed like something viable and similar to what I'd been researching in the past," he said. "I thought, 'Wow, we have something real.'"
Cannistra's increased interest began to encourage a growing number of students to research SDN, and the group began writing and developing modules for the OpenFlow Floodlight controller in their newly formed SDN certification lab. But even before the IBM grant, one of Cannistra's students developed a host-aware networking module, which became the start of a project that will bring SDN to the WAN and optical layer of the network.
Cannistra and his team have taken the module to Adva Optical Networking, a telecommunications vendor that provides equipment for data, storage, voice and video services. In turn, the module has helped Adva use OpenFlow SDN for WAN bandwidth provisioning. The company was the first to make SDN and programmability work at the optical layer by taking the OpenFlow commands from a Floodlight controller and converting them to Simple network Management Protocol, which talks to the data-flow plane on the optical layer.
"That [project] takes the dynamic networking concept a little further into the WAN," Cannistra explained. "We said, 'If we can do this on one data center and monitor the servers in one data center, why can't we take this concept and move it one step further into dynamic optical provisioning?'" This means managing it all the way down to the optical transport layer, and then do inter-data center connectivity, Cannistra said.
Projects abound in the SDN lab
Cannistra and his team continue to write and develop modules for the Floodlight OpenFlow Controller, thanks to the faculty grant from IBM. These days, Cannistra said his team is working on projects such as taking an OpenFlow network management application and rewriting it to incorporate more mobile devices. Avior, Cannistra said, is an open source, OpenFlow management application written for Floodlight by the SDN innovation lab at the college. The app was originally written to manage OpenFlow-capable devices, but Cannistra and his team rewrote it "with mobility in mind," he said.
"This means that there is no longer an installation required for the application to run properly. You can simply open a Web browser, visit the appropriate website, and the application is launched within the browser. … You're able to manage an OpenFlow controller from any device that has a Web browser."
Most recently, Cannistra and his team considered building a certification lab similar to the one at Indiana State University, which looks at OpenFlow standardization testing. However, Cannistra said he slowly started to realize that not only was that something he didn't want to do -- it was something that didn't jibe with Marist's approach to teaching. "At Marist we are innovators. We enjoy creating new technology and new uses for technology. I would much rather have our team work on new ways of using OpenFlow use cases and other SDN technologies. This fits into our mission as a college and our teaching philosophy," Cannistra said.
Instead, the college decided to forge ahead with an SDN innovation lab, which brings in different company partners and explores what can be done with SDN.
"We bring in different companies who want to do some innovative development and interoperability testing with other vendor equipment," he said. "It's more industry-focused. Students do academic research, but they also gain real-world exposure to companies who want to implement and develop [SDN], instead of having it sit out there as a theoretical project."
Marist is also partnering with other universities on projects stemming from the SDN innovation lab. "Columbia University has a wonderful optical-testing facility," Cannistra said. "So we'll be doing some work, they'll offset the optical engineering of it, and we'll do development on top of that." Cannistra and his team also started working with the City University of New York (CUNY), which is currently working on statistical analysis with VMware. "So [we're] taking that partnership and adding their work on top of our work," Cannistra said.
For example, CUNY is working on statistical analysis of VM migration time, Cannistra said. This research ties in with Marist's dynamic infrastructure test bed and development efforts. "So when they have their model completed, we will run their research model on our test bed to prove their theory," he said. "This will be a very powerful, effective and efficient addition to our work in our … SDN innovation lab. This is the type of collaboration that we all prosper from."
Bringing Floodlight into the OpenDaylight Project
In addition to the continual OpenFlow work, Cannistra and his team are beginning to "get engrossed with OpenDaylight development," he said. Cannistra pitched an idea recently at the SDN and OpenFlow World Congress in Germany to the chair of The OpenDaylight Project. "Hopefully, we'll be developing an open framework for OpenFlow development on the Floodlight controller," Cannistra said.
Researchers will begin determine how to port the Quality of Service module developed for Floodlight 1.0 into the OpenDaylight controller. "If you look at the OpenDaylight controller right now, it's not mobile-accessible. So we'll probably pitch another idea to do some mobile accessibility and development with the OpenDaylight controller."
Keeping SDN research, student-focused
Although Cannistra admitted there are "a lot of irons in the fire right now," the progress and partnerships Marist is making are essential to the student experience. "From my avenue, this is a wonderful technology, but I want to see students exceed," Cannistra said. "This is all student-driven. Sure, I'm the faculty behind this, but if it weren't for the students, this development wouldn't be done."
Through the various labs at Marist, students have gained invaluable knowledge and have made connections of their own during their time at the college. Cannistra's original lab assistant is now working on the core research team at EMC, while another student now works for Big Switch Networks, and a third works in the core networking group at Intel.
When looking at his time spent developing SDN and OpenFlow projects within the Marist labs, Cannistra said he's become more of an optimist in this space, as opposed to three years ago. "I would have been very skeptical," he said. "Right now, though, I see this as a close-knit group that's open for collaboration." This is particularly true with the OpenFlow protocol, he said.
Ideally, Cannistra wants to see the protocol be scalable, more robust and resilient on a production network.
"There are a lot of people backing this development with the OpenFlow protocol [now]. Do I think the end game is OpenFlow? No. I'm sure there are going to be new protocols that are developed over the years … but the concept of this protocol being open, accessible and providing intelligence from a central controller is pretty powerful."