Object database standards, persistence specifications, and physics data

Designers of data systems for next-generation physics experiments face a bewildering array of potential paths. On the one hand, object database technology is quite promising, and standardization efforts are underway in this arena. On another, lightweight object managers may offer greater potential for delivering the high performance needed from petabyte-scale data stores, and may allow more efficient tailoring to specific parallel and distributed environments, as well as to multilevel storage. Adding to the confusion is the evolution of persistence services specifications such as those promulgated by the Object Management Group (OMG). In this paper, we describe what we have learned in efforts to implement a physics data store using several of these technologies, including a lightweight object persistence manager aiming at plug-and-play with object databases, and a trial implementation of the OMG Persistent Data Services Direct Access protoco