I attended the Internet2’s Global Summit this week, an event that gathers a broad audience from Research and Education (R&E) networks from around the world, from engineers up to C-level attendees. The conference was jam-packed with excellent presentations, to which it’s impossible to do full justice. In this post, I’ll cover a couple of topics covered in the executive track on the first day of the conference.
One session was a panel consisting of Wendy Huntoon, President & CEO of Kinber, Ibrahim Ha, head of EMEA Infra Connectivity Programs at Facebook, and Chris Sedore, CEO of NYSERNET. A major theme of this session was more efficiently sharing resources. Wendy addressed the need to ensure that R&E leaders establish strong communications and onboarding for researchers to become aware of and get access to infrastructure resources and services. Without an easy way to onboard, researchers tend to give up on central IT teams and procure their infrastructure resources directly. Chris emphasized the need to cut down on needless duplication of investments such as fiber runs, particularly when looking at constrained resources. As an example, he mentioned that his fiber map looked remarkably similar to Internet2’s map in many places. Ibrahim addressed how Facebook overcome barriers to getting infrastructure to emerging countries in regions like Africa by collaborating with telecom and other webscale companies to share the costs of building subsea cables infrastructure to create access to bandwidth. One executive from AARNet (Australia’s version of Internet2), commented that R&E networks could gain tremendously from collaborating with Facebook and other webscale companies around fiber and other infrastructure-related topics.
In a separate session, Michael McCartney, the CIO of Purdue University’s main campus gave an insightful and humorous talk, telling the story of a successful IT as a Service transformation. When he started out, there were 67 distributed data centers on campus. He made it his mission to win researchers over to a new service model that utilized pre-planned scale-out purchases, one-time fees for what was in effect a five-year subscription, and a cloud-like model where unutilized capacity could be allocated on an as-needed basis to others. He made that model successful by hiring liaisons from academic backgrounds, who intimately understood researchers’ requirements and help them get what they needed. At first, folks on campus were more than skeptical, with one researcher telling Michael, “I’d rather remove my appendix with a spork than let you touch my computers.” But over time, he won them over, and the vast majority of the distributed data center owners happily moved to and stayed with the new services. Elias Eldayrie, CIO of the University of Florida told a very similar story in his talk, with similarly impressive results achieved. He recounted how UF’s central infrastructure services grew from supporting a tiny fraction of overall research spending to something like 75% over the course of a few years. In the context of a large research university that has $750M+ in annual research funding, that’s pretty impressive. Overall, both of their teams transformed central IT infrastructure services from minor to major players in supplying infrastructure to their campuses. In my next post on Global Summit, I’ll cover some nifty open source projects and how Kentik complements Internet2’s forthcoming DDoS scrubbing service.