Informatik-Kolloquium: Achieving in-memory performance with a disk-based system 7. Oktober 2020 | 17:00 Uhr - 18:00 Uhr
The increases in main-memory sizes over the last decade have made pure in-memory database systems feasible, and in-memory systems offer unprecedented performance. However, DRAM is still relatively expensive, and the growth of main-memory sizes has slowed down. In contrast, the prices for SSDs have fallen substantially in the last years, and their read bandwidth has increased to gigabytes per second. This makes it attractive to combine a large in-memory buffer with fast SSDs as storage devices, combining the excellent performance for the in-memory working set with the scalability of a disk-based system.
In this talk we present the Umbra system, an evolution of the pure in-memory HyPer system towards a disk-based, or rather SSD-based, system. We show that by introducing a novel low-overhead buffer manager with variable-size pages we can achieve comparable performance to an in-memory database system for the cached working set, while handling accesses to uncached data gracefully. We discuss the changes and techniques that were necessary to handle the out-of-memory case gracefully and with low overhead, offering insights into the design of a memory optimized disk-based system.
Thomas Neumann conducts research on database systems, focusing on query optimization (computing efficient query strategies) and query processing (efficient query execution). He studied business information systems at the University of Mannheim and received a doctorate in informatics from the same university in 2005. Before joining TUM (2010), Professor Neumann was a senior researcher at the Max Planck Institute for Informatics in Saarbrücken.
He acquired his postdoctoral teaching qualification (habilitation) in informatics from Saarland University (2010). He has been awarded the Gottfried Wilhelm Leibniz Prize of the German Research Foundation (2020).
Prof. Dr. Thomas Neumann, TU München
Zur Website der Veranstaltung →
Veranstaltung im Kalender speichern →