I build data-intensive systems that are more scalable, more robust, and more predictable. I draw from advanced statistical models to deliver practical database solutions to real-world problems. In particular, I adapt concepts and tools from applied statistics, optimization theory, and machine learning.
Barzan Mozafari is an Associate Professor of Computer Science and Engineering at the University of Michigan, Ann Arbor, where he leads a research group designing the next generation of scalable databases using advanced statistical models. Prior to that, he was a Postdoctoral Associate at MIT. He earned his Ph.D. in Computer Science from UCLA in 2011. His research career has led to several open-source projects, including BlinkDB (the first massively parallel approximate query engine), DBSeer (the first automated database diagnosis tool), and VerdictDB (the first platform-independent approximate query engine). He helped commercialize the ideas introduced by BlinkDB, as part of SnappyData, a company that was later acquired by TIBCO. He has won the National Science Foundation CAREER award, as well as several best paper awards in ACM SIGMOD and EuroSys.
MySQL, the most popular DB in the world, adopts our CATS algorithm as its the default scheduling strategy! Our scheduling algorithm, known as Contention-Aware Transaction Scheduling (CATS), is now the default policy in Oracle MySQL too! With this adoption, over 2M+ servers in the world are running CATS! Congratulations to my students, Jiamin Huang and Boyu Tian, for developing these algorithms! Read our papers in SIGMOD 2017, EuroSys 2017 and VLDB 2018.
Morris Wellman Faculty Development Assistant Professorship Humbled and honored to be named Morris Wellman Faculty Development Assistant Professor. Many thanks to Wellman family for their generosity.
CAREER Award: Designing a Predictable Database - An Overlooked Virtue Database research has mostly focused on improving the raw performance of database systems, while neglecting the predictability of their performance. NSF has funded us to rethink this traditional architecture, and build a new class of databases that guarantee predictability.
Vehicle-aware data management for autonomous cars Testing autonomous vehicles is an extremely involved task, as an enormous amount of data is constantly collected and processed by thousands of sensors. NSF has funded us to design a smart black-box for cars that can use effective data collection strategies for maximizing the likelihood of finding various types of faults!
Big Data Summer Institute in Bio-statistics Funded by National Institutes of Health, we have launched our Big Data Summer Institute, where you can learn about Big Data, statistics and bio-informatics while getting paid! All details and how to apply, can be found here (do NOT email me about this)!
Approximation for All.
DBSeer: Self-Driving Databases
automatic resource provisioning, automatic performance diagnosis, and automatic physical design.
Delivering sub-second latency when querying terabytes and petabytes of data.
Crowd-sourcing Big Data
Making crowd-sourcing scale up to tens of millions of tasks.
K*SQL and XSeq
High-performance complex event processing on complex data types such as RNA sequences, JSON files, and software traces.
Computer Science and Engineering
4769 Beyster Building
2260 Hayward St.
University of Michigan
Ann Arbor, MI 48109-2121
Office hours: by appointment only
|Nov 21||Google Kirkland|
|Nov 22||University of Washington|
|May 30||Teradata (San Diego)|
EECS 484: Database Management Systems
EECS 484: Database Management Systems
EECS 684: Current Topics in Databases
Our research is made possible through the generosity of the University of Michigan.
* This website's template is borrowed from Michael Bernstein and Jeffrey Heer.