Chandra’s Group Wins Best Paper at 2023 CCGrid Conference
Department of Computer Science & Engineering professor Abhishek Chandra’s research group won the best paper award at the 2023 International Symposium on Cluster, Cloud and Internet Computing (CCGrid). The conference was held in Bangalore, India, in early May.
The paper titled, “AggFirstJoin: Optimizing Geo-Distributed Joins using Aggregation-Based Transformations”, was led by University of Minnesota alumnus Dhruv Kumar (Ph.D., 2022) and Chandra, in collaboration with Sohaib Ahmad and Ramesh K. Sitaraman from the University of Massachusetts Amherst. Kumar was advised by Chandra during their Ph.D. program and is now an assistant professor at IIIT Delhi.
The paper focuses on the problem of geo-distributed data analytics. Data is increasingly being generated in an inherently distributed manner (from user devices, sites, and sensors across different locations). Traditionally, Internet companies and big businesses collect data and store it in a central location to run analysis, but this process is very costly and slow. By the time results are collected, that data may be outdated. Kumar and Chandra’s paper proposed a new method that makes this process more efficient, timely and cost effective for widely used database “join” queries that are typically very expensive.
“A lot of social media and internet services we use rely on how quickly we can analyze data and understand what is going on,” said Chandra. “For example, if you take a video streaming service like Netflix, it is important for them to understand which shows and movies users are watching at any given point in time. If one show is extremely popular, they need to know that so they can make sure that content is streamed efficiently to anyone who wants to see it without glitches. There are a number of applications for using this type of distributed data and our technique could help improve this process.”
Chandra’s group is working on a number of other projects in the distributed systems research area, including data collection for Internet of Things (IoT) devices, approximate data analytics, and implementing distributed systems in real-life settings.
“Broadly speaking, I am interested in taking data from users and devices and analyzing it quickly, efficiently and in a cost-effective manner,” said Chandra. “We are also working with Cisco on the idea of a smart classroom using edge computing to improve the experience for students. One example would be having lectures transcribed as they are happening or analyzing the whiteboard to get information to students in real time.”
CCGrid is a leading forum to disseminate and discuss research activities and results on a broad range of topics in distributed systems, ranging from computing clusters and high performance computing to widely distributed clouds and emerging Internet computing paradigms such as Fog and Edge Computing for Internet of Things (IoT) and Big Data applications. The conference features keynotes, technical presentations, posters, workshops, student symposium, industry interactions, the SCALE Challenge, and the co-located ICFEC conference.