One of the key bottlenecks in building machine learning systems is creating and managing the massive training datasets that today’s models learn from. In this talk, I will describe my work on data management systems that let users specify training datasets in higher-level, faster, and more flexible ways, leading to applications that can be built in hours or days, rather... Read More

Matt will present the following paper in this seminar: Title: Concurrent Prefix Recovery: Performing CPR on a Database Authors: Guna Prasaad, Badrish Chandramouli, Donald Kossmann Read More

This talk describes the Splice Machine Data Platform designed to power today’s new class of Operational AI applications that require high scalability and high-availability while simultaneously executing OLTP, OLAP and ML workloads. Splice Machine is a full ANSI SQL database that is ACID compliant, supports secondary indexes, constraints, triggers, and stored procedures. It uses a unique, distributed snapshot isolation algorithm... Read More

Tianyu will present this paper in this meeting: Title: Faster: A Concurrent Key-Value Store with In-Place Updates Authors: Badrish Chandramouli , Guna Prasaad , Donald Kossmann , Justin Levandoski , James Hunter , Mike Barnett Read More

Lin will present this work in this meeting: Title: Automatically Indexing Millions of Databases in Microsoft Azure SQL Database Authors: Sudipto Das, Miroslav Grbic, Igor Ilic, Isidora Jovandic, Andrija Jovanovic, Vivek R. Narasayya, Miodrag Radulovic, Maja Stikic, Gaoxiang Xu, Surajit Chaudhuri Read More

Prashanth will present the following paper in this meeting: Title: Thriving in the No Man’s Land between Compilers and Databases Authors: Holger Pirk, Jana Giceva, Peter Pietzuch Read More

Dana will present the following paper in this meeting: Title: Automated Performance Management for the Big Data Stack Authors: Anastasios Arvanitis, Shivnath Babu, Eric Chu, Adrian Popescu, Alkis Simitsis, Kevin Wilkinson Read More

Project 1: Storage Engine (Tianyu Li & Matt Butrovich) In this talk, we will discuss the work we've done on terrier's storage engine over the semester. We will cover the implementation of write-ahead logging and our proposed model for recovery, implementation of indexes, and our roadmap for the storage engine next semester. The immediate future direction for the storage work... Read More

Following from the idea that "one size no longer fits for all", a family of "NewSQL" specialized databases arose. To handle OLTP, researchers at MIT and Brown (and a few other places) built H-Store, a distributed, shared-nothing, in-memory database that got rid of locking, latching, buffering, and logging, beating the performance of traditional OLTP RDBMSs by nearly two orders of... Read More

n the last two decades, both researchers and vendors have built advisory tools to assist database administrators (DBAs) in various aspects of system tuning and physical design. Most of this previous work, however, is incomplete because they still require humans to make the final decisions about any changes to the database and are reactionary measures that fix problems after they... Read More