Over the course of the past decade, cloud-based networking has rapidly become the infrastructure underpinning our professional, social, and family lives. Even a small outage in the cloud can affect many people, leading to lost revenue, disruptions in connectivity, or security risks.
Enter Minlan Yu. An associate professor of computer science who specializes in data networking and cloud computing, she is using cutting-edge research to transform the way that telemetry—automated remote measurement of data—helps companies understand and manage their cloud-based networks. And, since she joined the School of Engineering & Applied Science in Yale’s Faculty of Arts and Sciences last year, she has been contributing her insights to the Department of Computer Science, to the new cross-disciplinary Computer Systems Lab @ Yale, and to the academic development of the undergraduate and graduate students who enroll in her courses and research in her lab.
At the heart of Yu’s efforts is the challenge of extracting useful information from a massive set of devices with limited resources. Today, in a typical operations center for a cloud-based network, a team of network administrators works round-the-clock observing a giant wall of screens that track data about the network’s activity. Even with experienced and highly-trained staff, this monitoring has its limits. Often the data available are only a sampling of overall network activity; to report on all data points all the time is resource-prohibitive. And simple human error can lead to a lapse in service that is costly or risk-inducing.
Yu focuses her research on developing new abstractions so that network operators can specify the data they want to monitor. This, in turn, enables her to design algorithms and system optimizations to collect the targeted data. She thus can provide insights into performance, failure, and security issues from the vast amount of data generated by the network. Her innovative approach is helping the tech world—from startups to Fortune 500 companies—save both expertise and money.
But Yu’s passion lies in academia, not in the admittedly lucrative corporate environment of Silicon Valley. Why? “It is a key difference,” she explained in an interview from her office in Arthur K. Watson Hall. “In industry, engineers are pushed to solve immediate problems. Academics can look more broadly. We can generalize. This is the great value of a university. We can provide fundamental solutions and bring longer-term benefits.”
She brings this philosophy—and her experience collaborating with well-known companies from Google to Microsoft to Facebook—to the classroom and to the lab. In her spring 2017 course, “Cloud Networking and Computing,” there was no required textbook. (“Books,” she said, “can’t keep up with the pace of change” in her field.) What students found instead was a dynamic analysis of real-world situations: of why these firms, whose products are so essential to our lives today, make the business and technology decisions they do. One class session might focus on tradeoffs, investigating why two companies faced with similar challenges might choose very different courses of action. In the next, Yu might stage a mini-competition in which groups of students try to build the fastest flow scheduling algorithms.
Yu speaks with excitement about the Computer Systems Lab, calling it “unique.” Founded last year, the lab sits at the intersection of computer science (CS) and electrical engineering (EE), with three EE faculty members and four in CS. This interdisciplinary configuration enables both faculty and students to work across traditional hardware and software boundaries.
With her faculty colleagues, Yu is taking part in a series of brainstorming lunches and planning more activities to leverage the combined talents of the lab’s members. And the graduate students in the lab are, Yu said, “a great group. They bring different strengths and new ideas.”
The development of the STEM pipeline has special meaning for Yu: she remembers well her own early forays into computer science. “From very young, as a little girl in China, I liked programming. I loved to see how I could use programming to draw pictures.”
Starting in the sixth grade, Yu participated in programming competitions and began to stand out in an arena largely dominated by boys. She always longed for opportunities to program together with more girls, and today she is doing just that, seizing every chance to encourage young women to pursue computer science. In this, she is following the example of her role model and former Ph.D. advisor, Jennifer Rexford, a computer scientist at Princeton.
In fact, it was on Rexford’s guidance that Yu first began to focus her research on issues of data centers and networking. It is a field of inquiry she came to “by chance”—one she is now illuminating for the next generation of Yale scholars. Together, they are finding solutions to the challenges of future technologies that most of us cannot yet even envision.
-Reported and written by Alison Coleman for the FAS Dean’s Office