Microsoft Research Puts Its Stamp on NSDI ’14
Among the many activities that occupy a research scientist, participation in conferences focused on an individual’s fields of interest ranks high. They represent an opportunity to meet with colleagues, get up to speed with what others are doing, and share some findings of your own.
Naturally, then, participants occasionally get asked to become organizers—an entirely different kind of sharing. Today, that’s where Ratul Mahajan of Microsoft Research finds himself.
Mahajan, a senior researcher in the Mobility and Networking Research Group, is serving as program co-chair for NSDI ’14, the 11th USENIX Symposium on Networked Systems Design and Implementation, which runs from April 2 through 4 in Seattle. So, a day before the event begins, he is simultaneously excited and keeping his fingers crossed.
“First and foremost,” he says, “I hope that we succeeded at picking an exciting program that advances the field. That is always the primary focus as program chair.”
It seems likely that he and fellow co-chair Ion Stoica of the University of California, Berkeley have accomplished just that. NSDI is the top-tier networked-systems conference, focusing on design principles, implementation, and practical evaluation of networked and distributed systems. The symposium aims to encourage a multifaceted approach to addressing a variety of research challenges.
The acceptance rate for NSDI is exclusive. Of 223 papers submitted, 38 have been accepted for this year’s event, seven of them from Microsoft Research, with its publications spanning a broad gamut of pertinent explorations.
“The Microsoft Research papers,” Mahajan says, “are a rich mix of technologies that touch the entire computing stack, from something as basic as localization to debugging problems in advertising systems.
“One mini-theme that stands out is that we are trying to do more with the cloud.”
Three of the papers from Microsoft Research address that topic. Blizzard: Fast, Cloud-scale Block Storage for Cloud-oblivious Applications—written by James Mickens, Edmund B. Nightingale, Jeremy Elson, and Darren Gehring of Microsoft Research; Bin Fan of Carnegie Mellon University; Asim Kadav and Vijay Chidambaram of the University of Wisconsin-Madison; and Osama Khan of Johns Hopkins University—discusses using the cloud as storage for PC applications.
Another, Bolt: Data Management for Connected Homes—by Trinabh Gupta of The University of Texas at Austin; Rayman Preet Singh of the University of Waterloo; and Amar Phanishayee, Jaeyeon Jung, and Mahajan of Microsoft Research—looks toward a future in which the Internet of Things has achieved broad acceptance.
A third cloud-related paper written solely by scientists from Microsoft Research—FaRM: Fast Remote Memory, written by Aleksandar Dragojević, Dushyanth Narayanan, Orion Hodson, and Miguel Castro—describes a main-memory distributed-computing platform that dramatically improves latency and throughput by using Remote Direct Memory Access rather than TCP/IP.
With NSDI being held for the first time in Seattle, Mahajan and Stoica have introduced an additional focus for the event.
“For this year’s symposium,” Mahajan says, “we wanted to set NSDI on the path toward greater research-industry interaction. We created the operational-systems track, intended for a different breed of papers.
“Rather than pure research results, the papers describe the design and experience with large-scale, operational systems and networks. They offer a behind-the-scenes look at real networked systems, which is otherwise hard to come by, especially for academic researchers. We hope this track will bring greater industry participation to the symposium and expose the research community to interesting problems being addressed by the industry.”
Two papers co-written by Microsoft Research scientists pertain to debugging complex systems:
- Adtributor: Revenue Debugging in Advertising Systems, by Ranjita Bhagwan, Rahul Kumar, Ramachandran Ramjee, and George Varghese of Microsoft Research, along with Surjyakanta Mohapatra, Hermanth Manoharan, and Piyush Shah of Microsoft.
- DECAF: Detecting and Characterizing Ad Fraud in Mobile Apps, by Bin Liu and Ramesh Govindan of the University of Southern California, along with Suman Nath and Jie Liu of Microsoft Research.
Other papers from Microsoft Research featured during NSDI ’14:
- cTPM: A Cloud TPM for Cross-Device Trusted Applications, by Chen Chen of Carnegie Mellon University, Himanshu Raj of Microsoft, and Stefan Saroiu and Alec Wolman of Microsoft Research.
- Epsilon: A Visible Light Based Positioning System, by Liqun Li, Guobin Shen, and Feng Zhao of Microsoft Research, Pan Hu from the University of Massachusetts Amherst, and Chunyi Peng of The Ohio State University.