Barzan Mozafari receives the EuroSys Test-of-Time award

Professor Barzan Mozafari, associate professor of computer science and engineering, and his co-authors have been honored with the 2023 Eurosys Test-of-Time Award in recognition of the sustained impact of their 2013 paper, “BlinkDB: Queries with Bounded Errors and Bounded Response Times on Very Large Data.”

EuroSys, the European chapter of the ACM Special Interest Group on Operating Systems (SIGOPS), honors just one to two papers a year with its prestigious Test-of-Time award. The award recognizes research published 10 years ago that has had a long-lasting and ongoing impact on the field, as deemed by the EuroSys Steering Committee and Officers.

Mozafari’s paper introduced BlinkDB, the first massively parallel approximate query engine for interactive SQL queries on large volumes of data. This engine was and remains innovative in that it allows users to adjust and balance the query engine’s accuracy and response time according to their needs. 

BlinkDB founded a new adaptive optimization framework to build and maintain different types of samples from the original data. It used a unique dynamic strategy to select appropriate samples at run-time based on the user’s accuracy and response time requirements. This allowed users to query as much as 17 terabytes of data in less than two seconds, which could be increased for users requiring more precise results. The result is a highly effective and adaptable SQL engine capable of running interactive queries over very large datasets.

Since its creation in 2013, BlinkDB has gone on to serve as a foundation for successive advances in query engines and data systems more broadly, enabling users to tailor their queries to better match their needs and extract useful information from very large datasets. The ideas introduced in this paper have since been widely adopted across the field and commercially, and it has been cited over 800 times. BlinkDB was later commercialized as SnappyData, and was acquired by TIBCO in 2019.

The co-authors of this paper are Samuel Ross Madden of the Massachusetts Institute of Technology, as well as Sameer Agarwal, Aurojit Panda, Henry Milner, and Ion Stoica of the University of California, Berkeley. The paper also earned the Best Paper Award at EuroSys 2013.