In 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. But these tools are 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 occur. What is needed is a truly "self-driving" database management system (DBMS) that is able to optimize the system for the current workload and predict future workload trends so that the system can prepare itself accordingly. It enables new optimizations that are not possible today because the complexity of managing these systems has surpassed the abilities of human experts. In this talk, I present Peloton, the first self-driving DBMS that we are building at CMU. Peloton's autonomic capabilities are now possible due to advancements in deep learning, as well as improvements in hardware and adaptive database architectures.
Dana Van Aken is a PhD student in Computer Science at Carnegie Mellon University advised by Dr. Andrew Pavlo. Her broad research interest is in database management systems. Her current work focuses on developing automatic techniques for tuning database management systems using machine learning.