Avatar

Let me share a dirty little secret about being a chief information officer in higher education: most of the decisions I get to make on a daily basis are not technical; they are economic. Given the rapid pace of change in higher education and the breadth of disruptive technologies—from massively open online courses, to consumer devices that change every eighteen months, to the latest “must have” student analytics tool—there are always more needs than resources, regardless of an institution’s size. CIOs prioritize needs against resources and make decisions on what priorities to pursue. In this blog, I want to argue that there are significant scientific and business reasons why research networking should be part of an institution’s portfolio of infrastructure investments.

Universities are the data-generation workhorses enabling modern scientific discoveries. The raw material for science, from the social sciences to the usual suspects in the STEM fields, is data. Data is growing exponentially, driven in parallel by increasingly sophisticated scientific instruments that generate petabytes of data and other devices capable of generating terabytes worth of data; these devices were once found only in core facilities but have now been commoditized and appear in multiple labs. The big data being generated by researchers doesn’t easily move across our enterprise campus networks, and it chokes when shackled with the modern firewalls that are necessary to protect our institutions.

This is why the Department of Energy developed the Science DMZ architecture and created a network enclave tuned for large data transfer that creates “fast lanes” on our networks. The first reason that research networking should be relevant to a higher education CIO is that science increasingly relies on the network. This dependency ranges from workflows that use the network to unite instruments with computational and storage resources to promote community insight, to the increasingly common first step of any experiment: grabbing data from a community repository. If a higher education CIO doesn’t pay attention to how the network can enable science, they risk putting their local research community at a competitive disadvantage.

I’m particularly excited to watch the development of Cisco’s Software Defined Access and Digital Network Architecture. This platform has the promise of creating more scalable and sustainable ways to support large data transfers and dynamically manage network resources, while also integrating more security than our research experiments have allowed.

The second and perhaps more practical reason for a higher education CIO to pay attention to research networking is because these efforts act as a “bird dog” for the next generation of challenges and opportunities that will drive the development of the enterprise network. Academic networking has a proud history of playing this role; the first “research network” that brought together the National Science Foundation’s Supercomputing Centers (called NSFnet) would—after the birth of the Web (another science-driven creation)—become the backbone of the modern Internet.

Current deployments of Science DMZs on campuses—and their growth across campus to support data-intensive science needs—right now sit in an awkward early-adopter phase. But, these research networking efforts are spurring the development of better software to manage large file transfers, and their accompanying failures; integrating tools like perfSonar and GridFTP into dashboards allowing us seamlessly to measure the network for benchmarks of real-world performance; and forcing us to acknowledge how complicated it is to identify the causes of poor data transfer performance (LAN, DTN, WAN?).

Perhaps the most important change happening now with research networking is not technical; it is cultural. Research networking requires us to get out of our network operation centers and into the labs of our user community. It moves our support conversations from “is the network working?” to “do you know how the network can make your work easier and faster?” Our engagement in research networking is changing the contextual expectations in the eyes of the end user. No longer is the network a dumb tube; it can now be a dynamic reconfigurable resource empowering work.

These transformations happen quickly, just look at the pace at which networking vendors are adopting innovations from research networking into their enterprise routing hardware. From software-defined networking being “baked” into modern routers to more plug-and-play support for equipment, and unified and simplified monitoring for these systems, what worked for research is increasingly the future of how work will be done across the enterprise.  In this regard, I’m particularly excited to watch the development of Cisco’s Software Defined Access and Digital Network Architecture. This platform has the promise of creating more scalable and sustainable ways to support large data transfers and dynamically manage network resources, while also integrating more security than our research experiments have allowed.

Put simply, the final reason to pay attention to research networking is that it is the easiest way to understand and experience tomorrow’s enterprise network.