Computer sciences' Akella makes networks 'future-proof'

December 17th 2014 Simon Kuran
Awards, Natural & Physical Sciences
Back to News

In our daily lives, we're surrounded by countless systems we typically take for granted. We expect lights to go on with the flick of a switch, and clean drinking water to flow with the twist of a faucet. And in the 21st century, we expect Internet access to be fast, reliable and always ready at the click of a mouse or swipe of a touchscreen.

These systems are the background of our lives, and most of us think about their operation only when something goes wrong. No news is good news, as they say — and an outage or persistent performance problem is bad news that can affect a large number of people and businesses.

Aditya Akella, an associate professor of computer sciences at the University of Wisconsin-Madison, is determined to reduce network failures and performance issues. Many of the key technologies he has invented operate within the core internals of network hardware and software today, keeping them happy and humming. His work may be somewhat opaque to the average Internet user, but its impact is far-reaching.

SIGCOMM, a subgroup of the Association for Computing Machinery (ACM) devoted to communications and computer networks, honored Akella with its 2014 Rising Star Award. ACM is the world's leading professional organization in computing.

Akella flew to Sydney, Australia, to accept the award in early December at the CoNEXT 2014 conference, where he also delivered the keynote speech.

Making networks "future-proof" is a leading goal for Akella. And it's a tall order in a rapidly changing tech environment marked by a profusion of devices and computing that is embedded into almost every facet of life.

"I want to make networks perform well and be robust, even as new, disruptive technologies arise, including both hardware and software," says Akella.

Akella credits UW-Madison's collaborative environment as a major factor in his success. "Being at UW has been a great asset," he says. "I couldn't replicate this work anywhere else."

"Being at UW has been a great asset. I couldn't replicate this work anywhere else."
— Aditya Akella

Colleagues within the Department of Computer Sciences and the Division of Information Technology have provided access to data on things like network configuration and traffic patterns that helped Akella design suitable systems and validate their performance. "Such data is key to building a research program grounded in reality, and to having real-world impact," he says.

Modeling network complexity is one of Akella's key innovations. Complexity is what makes networks failure-prone, he says. By analyzing what makes them problematic, robustness can be improved. To that end, Akella developed the first complexity metrics for network design, helping to pinpoint the likelihood of a particular network's failure and what factors might contribute to a breakdown.

This type of complex, high-impact work in network management is one reason Akella was honored as a SIGCOMM Rising Star. His research in content distribution is another. It's another element of our digital lives that is hidden from view, yet critical.

Akella helps enable the smooth delivery of Internet content, and thereby improves users' quality of experience, through what are called redundancy elimination technologies. Rather than deliver all the bytes a user requests, these systems instead provide a pointer to the most frequently requested and essential information.

"No matter what new things arise — like new applications and workloads — certain information is popular," he says. "There is repetition across time and space, and it's about trying to ensure the optimal delivery of such content. The idea is to be resource-efficient and fast by removing redundant bytes and speedily delivering the most useful bytes."

Akella's strategies help not only with improving user-perceived performance, but also with energy and battery constraints and bandwidth costs. A redundancy-elimination system he developed called EndRE is part of Microsoft's Windows 8 operating system. Other systems he's developed have been adopted by Dell, and some are under trial for adoption at the Bay Area startup Instart Logic.

UW-Madison has proven to be a fitting academic home for the award-winning researcher, who joined the faculty in 2006. Among his current endeavors, Akella is the co-principal investigator of the National Science Foundation-funded CloudLab project, which will explore future cloud architectures and also support high-level research in numerous disciplines.

"We are unique here," he says, citing UW-Madison strengths in systems, networking and cloud computing. "We lead the world in large infrastructure research and education."

Story by Jennifer A. Smith, Department of Computer Sciences