An Approach for Data Access Using CORBA and Objectivity
The software architecture being developed by the High Energy and Nuclear Physics Grand Challenge (HENP-GC) group attempts to address data access problems that arise from multi-user explorations of very large-scale data samples. Such explorations will become common in the coming years as new high energy and nuclear physics experiments begin producing immense volumes of data on a yearly basis. The detectors being built at Brookhaven’s Relativistic Heavy-Ion Collider (RHIC) are among those that will produce such very large data volumes. The HENP-GC project is working to provide tools to the RHIC Computing Facility (RCF) for managing the analysis of these massive data sets.
In its current incarnation, the HENP-GC architecture uses Common Object Request Broker Architecture (CORBA) for communication between the software components: a query estimator, a cache manager, a query monitor, and (clients within) the users’ analysis codes. Object management is handled with Objectivity. Accomplishing the current integration has required addressing a number of design and implementation issues. Design features such as how to efficiently pass large collections among CORBA components have implications on both the system performance and the component dependency upon knowledge of the underlying database structures. This talk describes the experience of the HENP-GC project in developing software for a mixed CORBA/Objectivity environment. It will cover a number of design issues that the project has addressed and discusses some of the performance implications of the adopted approach.