Jump label

Service navigation

Main navigation

You are here:

Main content

Research

We're the database group at TU Dortmund. We design, build, and evaluate database systems, ready to handle large amounts of data fast, efficient, and safely.

Focus of our research is the implementation of databases on modern hardware architectures. We're pioneers in the use of field-programmable gate arrays (FPGAs) for database acceleration. We showed how modern network cards with hardware acceleration (by means of remote direct memory access, RDMA) can be leveraged for distributed data processing. And we also work on techniques for in-memory database processing on modern multi-core hardware.

  • MxKernel: A Bare-Metal Runtime System for Database Operations on Heterogeneous Many-Core Hardware
    The MxKernel project is part of the DFG Priority Program 2037 “Scalable Data Management for Future Hardware”. MxKernel is a joint project with Olaf Spinczyk's Embedded System Software Group at TU Dortmund University. MxKernel's theme is Database/Operating System Co-Design. With MxKernel, we want to build a new bare-metal runtime system. MxKernel provides very lightweight resource management for database and operating system, which both run as equal peers on top of the runtime. In MxKernel, heterogeneity and parallelism become first-class citizens, ready also for memory hierarchies or modern storage technologies (such as non-volatile memories). Instead of a classical "thread" model, MxKernel provides MxTasks as an abstraction for work items.
  • RAPP Center: Ruhr Astroparticle and Plasma Physics Center
    The DBIS Group participates in the Ruhr Astroparticle and Plasma Physics Center (RAPP Center), a joint effort between Ruhr-Universität Bochum, TU Dortmund University, and Universität Duisburg-Essen.
  • Energy Awareness in Database Algorithms and Systems (SFB 876, A2)
    Energy consumption has become a key factor in modern computing systems. The behavior of algorithms and programs may have important effects on the energy consumption characteristics of a machine. As part of the sub-project A2: Algorithmic Aspects of Learning Methods in Embedded Systems of SFB 876, we investigate how we can improve energy efficiency by making data-intensive algorithms more energy aware.
  • Real-Time Analysis and Storage for High-Volume Data in Particle Physics (SFB 876, C5)
    Sub-project C5: Realtime Analysis and Storage for High-Volume Data in Particle Physics of SFB 876 addresses the very large data volumes that arise in scientific applications such as from the LHC particle accelerator at CERN. We want to develop techniques to allow to pre-filter the data voume of approximate 30 gigabytes per second, or 400 PB per year, but also persistently store and analyze the remaining data (about 20 PB per year) using distributed storage and analysis techniques.
  • In-Memory Joins
    Together with researchers from ETH Zürich and U Waterloo, we are developing and evaluating strategies to implement database algorithms on modern hardware architectures. They are geared toward massively parallel in-memory processing. We currently hold the world record in join processing in such environments.
  • Avalanche - Stream Processing on Bare Metal
    The Avalanche project pioneered the use of FPGAs for database acceleration. “Glacier” is a query compiler that translates a rich subset of SQL into equivalent hardware circuits. This allows stream processing at guaranteed network speed and thus removes a bottleneck that had hampered many important application classes (such as high-frequency trading, HFT) before.
  • Data Cyclotron: Data Processing on Hardware-Accelerated Networks
    The speed of modern network standards has long surpassed hard disk bandwidths and approaches the speed of machine-internal interconnect technologies (memory bus, PCIe, etc.). This suggests to put the ancient wisdom of distributed database systems–try to avoid communication at (almost) any cost–upside-down. In Data Cyclotron, a joint project with the database group at CWI in Amsterdam, the strategy is to leverage the available bandwidth, rather than spending efforts in trying to avoid it.
  • Pathfinder: XQuery Compilation Techniques for Relational Database Targets
    With Pathfinder we designed an XQuery compiler that brings together the performance of relational engines and the expressiveness of XQuery. Pathfinder shreds XML document into a relational format (using the XPath Accelerator technique), leverages advanced XPath evaluation techniques on such encoded data (such as staircase join), and compiles XQuery into purely relational query plans that fit the strengths of relational execution back-ends. The technology built into Pathfinder can be combined with any relational database back-end, which we demonstrated with an SQL code generator for the Pathfinder compiler. The primary target of the Pathfinder compiler is, however, the main-memory database engine MonetDB. The combination MonetDB/Pathfinder is also distributed as the open source engine MonetDB/XQuery.


Sub content

Contact

Prof. Dr. Jens Teubner
Tel.: 0231 755-6481