Hybrid2-Index Structures for Main Memory Databases
Funded by
DFG
Runtime:
2019-2023
Project Coordinator:
Research Associate:
Motivation
Because index access is the starting point for subsequent processing steps of queries in DBMS, fast index access is essential fot the total performance of a DBMS. If data acess would be done without an index, the complete dataset have to be read evan if we need only a small part of the dataset.
Goals and Procedure
In this DFG-project efficient indexstructures for main memory databases on hardware accelerators should be researched. This project will be worked on in conjunction with the University of Magdeburg (Prof. Thilo Pionteck), which will focus on index structure on FPGAs and the University of Lübeck will focus on acceleration of index structures through graphic processing units (GPU).
Innovation and Prospects
We research and implement different data structures for key-value based index stores and use different hybrid systems with CPU, GPU and APU accelerators to achieve higher performance than in none hybrid scenarios.
Project Partners
Prof. Dr.-Ing. Thilo Pionteck, Institute for Information Technology and Communications, Otto von Guericke University of Magdeburg
Activities
- The 6th International Conference on Applied Machine Learning and Data Analytics (AMLDA) 2023 in Lübeck, 9.11.2023 (online) and 10.11.2023 (hybrid), General Chairs: M. A. Jabbar, Fernando Ortiz Rodriguez, Sanju Tiwari, Sven Groppe.
- International Workshop on Big Data in Emergent Distributed Environments (BiDEDE 2023) in conjunction with ACM SIGMOD/PODS in Seattle, WA, USA: June 18, 2023, Workshop Chairs: Sven Groppe, Le Gruenwald, Ching-Hsien Hsu
- "Norman Revell Best Paper Award" for the publication "Tobias Groth, Sven Groppe, Thilo Pionteck, Martin Koppehel, Franz Valdiek, Accelerated Parallel Hybrid GPU/CPU Hash Table Queries with String Keys, The 33rd International Conference on Database and Expert Systems Applications (DEXA), Vienna, Austria, https://doi.org/10.1007/978-3-031-12426-6_15", 22.8.2022
- International Workshop on Big Data in Emergent Distributed Environments (BiDEDE 2022) in conjunction with ACM SIGMOD/PODS in Philadelphia, PA, USA: June 12, 2022, Workshop Chairs: Sven Groppe, Le Gruenwald, Ching-Hsien Hsu
- Bachelor-/Master-Forum 2021, "Utilizing Emergent Technologies for Big Data Analytics" (Folien: HTML, PDF), Universität zu Lübeck (online), Prof. Dr. Sven Groppe, 15.12.2021
- Workshop together with DIMA (TU Berlin), online, "Luposdate3000: Towards (Semantic) Hybrid Multi-model Multi-platform ((S)HM3P) Databases" (HTML, PDF), Prof. Dr. Sven Groppe, "IoT-DB-Simulator and Combining Query Processing with RPL-Routing" (PDF), Benjamin Warnke, 21.10.2021
- International Workshop on Big Data in Emergent Distributed Environments (BiDEDE 2021) in conjunction with ACM SIGMOD/PODS in Xi'an, Shaanxi, China: June 20-25, 2021, Workshop Chairs: Sven Groppe, Le Gruenwald, Ching-Hsien Hsu
- Keynote on the International Semantic Intelligence Conference (ISIC 2021), New Delhi (hybrid), India: February 25, 2021 with the title "Emergent Hardware for Databases with focus on Semantic Web: Semantic Hybrid Multi-model Multi-platform (HM3P) Databases" (HTML, PDF): Prof. Dr. Sven Groppe
- Talk in International Federation for Information Processing (IFIP) Working Group WG2.6 Database seminar: Hybrid Multi-Model Multi-Platform (HM3P) Databases (HTML, PDF), February 17, 2021, Prof. Dr. Sven Groppe
Publications
Hybrid CPU/GPU/APU accelerated query, insert, update and erase operations in hash tables with string keys
Knowledge and Information Systems, 65 (10) :4359--4377
2023
Herausgeber: Springer Science and Business Media {LLC}
Datei: | s10115-023-01891-w |
Bibtex: | @article{groth2023hybrid, title={Hybrid CPU/GPU/APU accelerated query, insert, update and erase operations in hash tables with string keys}, author={Groth, Tobias and Groppe, Sven and Pionteck, Thilo and Valdiek, Franz and Koppehel, Martin}, journal={Knowledge and Information Systems}, pages={4359--4377}, year={2023}, publisher={Springer} } |
Accelerated Parallel Hybrid GPU/CPU Hash Table Queries with String Keys
The 33rd International Conference on Database and Expert Systems Applications (DEXA), Vienna, Austria
2022
Datei: | 978-3-031-12426-6_15 |
Bibtex: | @inproceedings{Groth2022GPUTable, title = {Accelerated Parallel Hybrid GPU/CPU Hash Table Queries with String Keys}, author = {Tobias Groth and Sven Groppe and Thilo Pionteck and Martin Koppehel and Franz Valdiek}, booktitle = {The 33rd International Conference on Database and Expert Systems Applications (DEXA), Vienna, Austria}, url = {https://doi.org/10.1007/978-3-031-12426-6_15}, year = {2022}, note = {This paper received the Norman Revell Best Paper Award} } |
CuART - a CUDA-Based, Scalable Radix-Tree Lookup and Update Engine
50th International Conference on Parallel Processing (ICPP 2021), Lemont, IL, USA
Herausgeber: Association for Computing Machinery (ACM)
2021
Datei: | 3472456.3472511 |
Bibtex: | @inproceedings{Koppehel21CuART, author = {Martin Koppehel and Tobias Groth and Sven Groppe and Thilo Pionteck}, title = {CuART - a CUDA-Based, Scalable Radix-Tree Lookup and Update Engine}, year = {2021}, publisher = {Association for Computing Machinery (ACM)}, url = {https://doi.org/10.1145/3472456.3472511}, booktitle = {50th International Conference on Parallel Processing (ICPP 2021)}, location = {Lemont, IL, USA} } |
Parallelizing Approximate Search on Adaptive Radix Trees
Proceedings of the 28th Italian Symposium on Advanced Database Systems, Villasimius, Sud Sardegna, Italy (virtual due to Covid-19 pandemic), Seite 56--67
2020
Datei: | 16-paper.pdf |
Bibtex: | @inproceedings{GrothGKP20, author = {Tobias Groth and Sven Groppe and Martin Koppehel and Thilo Pionteck}, title = {Parallelizing Approximate Search on Adaptive Radix Trees}, booktitle = {Proceedings of the 28th Italian Symposium on Advanced Database Systems, Villasimius, Sud Sardegna, Italy (virtual due to Covid-19 pandemic)}, pages = {56--67}, year = {2020}, url = {http://ceur-ws.org/Vol-2646/16-paper.pdf} } |