Distributed Data Management
The increasing use of high-volume scientific data requires new methods of distributed data management. Replication and synchronization, consistency and fault tolerance are the core research fields of this working group. We implement our theoretical results in practical data storage systems based on structured peer-to-peer overlays and distributed file systems.
In cooperation with other scientific disciplines and industry we improve, review and adapt our solutions (Scalaris, XtreemFS, BabuDB, etc.) in the context of real-world applications. Our research topics include: transactional key/value stores (Scalaris), distributed file systems (XtreemFS), and logstructured matadata catalogs (BabuDB).