As mentioned above, the current crop of IRSA services (http://irsa.ipac.caltech.edu/) is purely HTML/CGI-based. While perforce confined to the server, the same architectural approach is used of using a collection of loosely-coupled tools to quickly build an array of user services (some generic and some of very narrow applicability). Here the components are a set of stand-alone executables (mostly C code) which have been constructed to allow command/control interaction with a driver application (the CGI program in most cases).

By avoiding any dependence on a software "system" infrastructure and the single-interface problems outlined earlier, we find that we can address the real specific needs of our user community much more effectively and rapidly.

It is in part this operational experience which convinces us of the validity of a loosely-coupled, "systemless" approach. Indeed, many of these components on the server side have simply been wrapped to provide the remote data search and access services that are part of our JAVA-based system.

Related Activities

As mentioned earlier, much of the IRSA effort goes toward effectively managing the data within the DBMS itself. Besides basic data ingestion and database administration, this involves extending the database server software to allow more efficient searching algorithms to be applied to the data. This work is funded through various IPAC projects as well as directly by NASA.

IRSA also manages an NSF supercomputer project call the Digital Sky through a research grant from the National Partnership for Advanced Computing Infrastructure (NPACI). Digital Sky efforts are focussed on the problem of interarchive collaboration, intercomparing and providing the equivalent of database "join" capabilities (based primarily of fuzzy positional criteria) for very large tables widely distributed across heterogeneous DBMSs and platforms.