Interfacing to distributed data archives

Abstract

The general problem of managing large data archives or libraries of digital data is particularly challenging when the system must cope with active data which is processed on-demand. Conventional data archives are designed so that data is ingested into the system, and retrieved by users to satisfy particular requests. An active data archive can be defined as one where much of the data is generated on-demand, as value-added data products or services derived from existing data holdings. Interfacing users and applications to such an archive system requires a more complex infrastructure than conventional or passive archive systems. We describe two archive systems we have built for providing web-based access to satellite and geospatial imagery as well as to medical imagery such as that from the Visible Human datasets. We contrast the requirements and features of these archives and discuss the Java and CORBA software infrastructure that we have developed to interface to them. We also describe our integration of commercial products, such as StudioCentral and Informix with our software and some of the geospatial standard interface definitions, such as GIAS and OpenGIS.