In contrast to a typical introductory course in database systems where one learns to design and query relational databases, the goal of this course is to get a fundamental insight into the implementation aspects of systems designed to manage and process large amounts of data. Our objective in this respect is two-fold. (1) To gain the background required to design and implement future data management and processing systems and (2) to gain an understanding of how performance of practical data management systems can be tweaked.
In particular, we take a look under the hood of relational database management systems, with a focus on query and transaction processing. The focus on relational database management systems is motivated by the fact that the algorithms and architectures underlying relational databases have strongly influenced the design of contemporary data processing and management systems: graph databases, in-memory database systems, stream databases, and even NoSQL systems.
With respect to query processing, we study the whole workflow of how a typical relational database management system optimizes and executes SQL queries. This entails an in-depth study of: (1) translating the SQL query into a “logical query plan”; (2) optimizing the logical query plan; (3) how each logical operator can be algorithmically implemented on the physical (disk) level, and how secondary-memory index structures can be used to speed up these algorithms; and (4) the translation of the logical query plan into a physical query plan using cost-based plan estimation.
With respect to transaction processing we study how a typical relational database management systems ensures recovery from errors and controls concurrent access to the data. Topics studied in transaction processing include logging, serializability, concurrency control, and their combination.
For the table of contents, course notes, slides, exercises and solutions, as well as recording of the lectures, see the Virtual University page.