To whom it may concern: I am Head of ePublishing and eScience atFIZ Karlsruhe, a non-for-profit company providing services for knowledge management and information transfer to the STM and academic community. My current activities include the project eSciDoc, which aims to build an e-research infrastructure for the Max Planck Society, and the promotion of the project results to a broader community in Germany and abroad. His other interests include building online systems for commercial publishers, collaboration tools for scholars and scientists, and information management for academia and research.
For my man project eSciDoc, we have done an extensive research on repository systems - back in 2005. At that time, Fedora had a unique and outstanding feature set. I wuld say this is still true, but the Fedora team worked hard to improve both reliability and performance. At FIZ Karlsruhe, we see Fedora as one of our strategic software platforms for future projects. For one of those, we did some research on scalability and ingest performance of Fedora 3. If you are interested in the (preliminary) results, check out our wiki.