1.
Dreseler, M., Kossmann, J., Boissier, M., Klauck, S., Uflacker, M., Plattner, H.: Hyrise Re-engineered: An Extensible Database System for Research in Relational In-Memory Data Management. 22nd International Conference on Extending Database Technology (EDBT). pp. 313–324 (2019).
Research in data management profits when the performance evaluation is based not only on individual components in isolation, but uses an actual DBMS end-to-end. Facilitating the integration and benchmarking of new concepts within a DBMS requires a simple setup process, well-documented code, and the possibility to execute both standard and custom benchmarks without tedious preparation. Fulfilling these requirements also makes it easy to reproduce the results later on. The relational open-source database Hyrise (VLDB, 2010) was presented to make the case for hybrid row- and column-format data storage. Since then, it has evolved from being a single- purpose research DBMS towards becoming a platform for various projects, including research in the areas of indexing, data partitioning, and non-volatile memory. With a growing diversity of topics, we have found that the original code base grew to a point where new experimentation became unnecessarily difficult. Over the last two years, we have re-written Hyrise from scratch and built an extensible multi-purpose research DBMS that can serve as an easy-to-extend platform for a variety of experiments and prototyping in database research. In this paper, we discuss how our learnings from the previous version of Hyrise have influenced our re-write. We describe the new architecture of Hyrise and highlight the main components. Afterwards, we show how our extensible plugin architecture facilitates research on diverse DBMS-related aspects without compromising the architectural tidiness of the code. In a first performance evaluation, we show that the execution time of most TPC-H queries is competitive to that of other research databases.
2.
Kossmann, J., Schlosser, R.: A Framework for Self-Managing Database Systems. 2019 IEEE 35th International Conference on Data Engineering Workshops (ICDEW). pp. 100–106 (2019).
Database systems that autonomously manage their configuration and physical database design face numerous challenges: They need to anticipate future workloads, find satisfactory and robust configurations efficiently, and learn from recent actions. We describe a component-based framework for self-managed database systems to facilitate development and database integration with low overhead by relying on a clear separation of concerns. Our framework results in exchangeable and reusable components, which simplify experiments and promote further research. Furthermore, we propose an LP-based algorithm to find an efficient order to tune multiple dependent features in a recursive way.
3.
Schlosser, R., Kossmann, J., Boissier, M.: Efficient Scalable Multi-Attribute Index Selection Using Recursive Strategies. 35th IEEE International Conference on Data Engineering, ICDE. pp. 1238–1249. IEEE (2019).
An efficient selection of indexes is indispensable for database performance. For large problem instances with hundreds of tables, existing approaches are not suitable: They either exhibit prohibitive runtimes or yield far from optimal index configurations by strongly limiting the set of index candidates or not handling index interaction explicitly. We introduce a novel recursive strategy that does not exclude index candidates in advance and effectively accounts for index interaction. Using large real-world workloads, we demonstrate the applicability of our approach. Further, we evaluate our solution end to end with a commercial database system using a reproducible setup. We show that our solutions are near-optimal for small index selection problems. For larger problems, our strategy outperforms state-of-the-art approaches in both scalability and solution quality.