Software-defined networking (SDN) has its roots anchored deeply in education, and its impact has been largely on the academic side, where technologies like OpenFlow have allowed researchers to open up networking operating systems for those looking to program the network and alter forwarding behavior.
But it turns out that universities do more than research; they also serve a diverse set of users with varying needs. This requires them to build full campus networks in much the same way as enterprises do. As higher education grapples with the same drivers as the rest of the world -- more users accessing the network from more places and driving additional traffic -- it might find that SDN is the key to building more capable networks.
In many ways, university networks experience the same stressors that drive network architects mad in the enterprise world. They often serve large user bases who are physically distributed and have a healthy appetite for bandwidth. These users are fans of streaming content (Netflix and YouTube are notoriously popular in the dorms), and they make use of application types (big data apps like Hadoop) that you might see in more commercial industries like AdTech or Search. But universities arguably have it tougher.
The diurnal problem
They have one user base active during the day and another active at night. The colleges that house research teams require high-bandwidth, low-latency connectivity for data sharing, both within the university confines and across campuses. At night, the user base shifts as students leave class and return to their rooms to continue their work, consume content, and play games.
At the most basic level, the diurnal patterns of network utilization mean that universities are building networks that behave one way during the day and another at night, prompting the question: How do you build a network with varying requirements?
It starts with capacity. Many universities have actually prepared for their bandwidth-hungry users by bringing in more fiber than most would expect. Fiber-optic connectivity lends itself well to the task of connecting various buildings across a sprawling geography. This leaves many campuses more prepared for the bandwidth explosion that enterprises have been experiencing for years.
But the real issue is that the connectivity requirements shift from the college labs to the dorms when the population heads home for the night.
SDN provides a measure of control that allows campuses to deal with their shifting capacity needs. Combining SDN with the underlying optical transport allows university architects to cable up a campus statically but program paths dynamically. Using technologies like wavelength division multiplexing (WDM), network architects can provision paths through the network that vary as the user load shifts. During the day, capacity can be allocated to the researchers. At night, that capacity can be dialed down to allow students more access to content and games.
In a statically provisioned world, this is labor intensive and prone to error, but SDN provides a central point of control, typically with programmatic APIs that allow network engineers to automate capacity allocation.
Of course, the shift from labs to dorms is never absolute. Universities have scarce resources in many of their labs, which leaves researchers grabbing lab time whenever it is free. Even when the bulk of users have moved from the campus to their dorms, there are still applications that have to be supported. So who gets priority: education or entertainment?
It might be as simple as giving priority to anything originating from the campus. But remote access means that researchers might be working from their rooms, and wireless connectivity means that gamers might be playing from the campus. The real challenge is not about location as much as it is about applications. Universities must be able to identify applications and throttle the performance up or down based on what is most important. SDN allows networks to become more application aware and then adjust the experience dynamically.
The challenges for universities extend further still. How do you provide security for a vast array of projects that include everything from individual student pet projects to massive collaborations with other universities? The strategy for most is to create a demilitarized zone (DMZ), hosting those applications that have to be secure behind a foreboding perimeter and pushing everything else to an open environment.
But this creates problems in and of itself. As a user, being inside the zone means you are protected, but you pay the price of requiring all changes to be approved by someone else. Research is, by its very nature, an exercise in iteration, and these delays are not always palatable. This is why shadow IT is on the rise in many campus environments. However, choosing agility at the expense of security isn't always practical, either. What if your project is related to statistics coming from a learning hospital, which imposes HIPAA restrictions on data access?
SDN allows network teams to provide the type of control necessary to support change without necessarily relegating resources to the Wild, Wild West outside the perimeter. For example, network engineers can leverage their rich fiber environment and allocate wavelengths as a service to individual research teams. By isolating traffic on a wavelength, they guarantee bandwidth and ensure application isolation. What is the glue that makes this type of management possible without creating IT resource bottlenecks? SDN.
Ultimately, SDN isn't a panacea. But using SDN in fiber-rich environments can make a difference for universities looking to benefit from technology -- especially one they had a heavy hand in creating.