Increasing volumes of data, varying workloads, and complex systems make database administration increasingly challenging for human database administrators. Autonomous or self-driving database systems utilize their knowledge of processed workloads, the stored data, and other runtime information to support database administrators in their tasks or to optimize the system's configuration without any human intervention. For example, such systems are capable of selecting indexes that substantially improve the system's runtime performance.
There are various challenges regarding autonomous approaches: (i) achieving robust and efficient optimization by relying on heuristics, optimization, or machine learning methods; (ii) integrating such approaches into database systems with acceptable implementation and runtime overhead; and (iii) mitigating trust issues of database administrators and users that are caused by non-explainable decisions made by autonomous systems.
There are several potential topics for master's theses available in the areas above.
Contact: Jan Kossmann