Efficient Architectures for Data-Intensive Computing

The Chair of High-Performance Computing, led by Prof. Dr.-Ing. André Brinkmann, focuses on the development of scalable algorithms and systems for high-performance computing and cloud architectures. Central aspects of the research are the design of efficient storage systems and the management of large-scale data.

Research topics include parallel file systems such as Lustre and GekkoFS, emerging network technologies like Data Path Processors (DPUs), and compression techniques such as deduplication and delta encoding. Additional areas of interest are modern storage technologies (SSDs and persistent memory) along with suitable indexing structures.

Further work addresses performance analysis in HPC environments, the development of trace archives, and load balancing for both traditional and malleable systems, with the goal of sustainably improving the efficiency and scalability of future computing systems.

Most recent publications

  • J. Ostermann, F. Schimmelpfennig, M. Maesker, A. Brinkmann, und R. Salkhordeh, "HyperionR - Experiences transforming an optimized Trie index data structure from C to Rust" in Proc. 39th GI/ITG International Conference on Architecture of Computing Systems (ARCS), March 24-26, Mainz, Germany, 2026.
  • F. Schimmelpfennig, J. Sass, R. Salkhordeh, M. Kröning, S. Lankes, und A. Brinkmann, "DPA-Store: An Ordered Network Data Path Key-Value Store" in Proc. 20th USENIX Symposium on Operating Systems Design and Implementation (OSDI), July 13-15, Seattle, WA, USA, accepted for publication, 2026.
  • A. Sedaghatgoo, R. Salkhordeh, A. Brinkmann, und H. Asadi, "Towards Scalable Storage Architectures for GPU Clusters Running Large Language Models" in Proc. ACM SIGMETRICS, Ann Arbor, Michigan, USA, June 8-12, accepted for publication, 2026.
  • J. Sass, A. Brinkmann, M. Bjørling, X. He, und R. Salkhordeh, "ZTL: A block layer ZNS driver" J. Syst. Archit. vol. 175, p. 103757. 2026.
    doi: 10.1016/J.SYSARC.2026.103757