Automating Cloud Databases with Interpretable Machine Learning

General public Colloquium / Congress / Forum

Database management system (DBMS) automation is an indispensable component of practical and performant databases systems. Over the last several decades, academia and industry built a myriad of DBMS automation tools to assist database administrators. Despite significant progress, the advent of cloud computing turned the automation problem into a constant battle of modeling ever-changing cloud hardware and software while meeting the customer/workload needs.

In the first part of this talk, we give an overview of MySQL HeatWave, which is an Oracle Cloud service that combines MySQL, the industry-proven and world’s most popular open-source database, with a massively parallel, in-memory query engine, HeatWave, that can execute queries over large scale data by speed-ups ranging from 100x to 1000x compared to native MySQL.
In the second part of the talk, we discuss MySQL Autopilot, which integrates a diverse set of ML-based automation capabilities into MySQL HeatWave. These ML-based capabilities are designed to simplify management and automatically adjust performance/cost based on workloads. We demonstrate various MySQL Autopilot use cases and conclude the talk with research agenda and open problems.

When? 15.12.2021 16:15
Where? PER 21 G120
Bd de Pérolles 90
1700 Fribourg
speaker Dr. Onur Kocberber, Oracle Labs, Zurich
Contact Département d'Informatique
Stéphanie Fasel
Bd de Pérolles 90
1700 Fribourg
0263008322 90