The next generation of telescopes, such as the Square Kilometre Array (SKA), will generate orders of magnitude
more data than previous instruments, far in excess of current storage and networking system handling abilities.
To address this problem, we propose an architecture where data is distributed over several archive sites, each
holding only a portion of the overall data, that provides efficient and transparent access to the archive as a whole.
This paper describes that architecture in detail and the design and implementation of a prototype system, based
on the Integrated Rule-Oriented Data System (iRODS) software.
The ALMA Common Software (ACS) provides both an application framework and CORBA-based middleware
for the distributed software system of the Atacama Large Millimeter Array. Building upon open-source tools
such as the JacORB, TAO and OmniORB ORBs, ACS supports the development of component-based software in
any of three languages: Java, C++ and Python. Now in its seventh major release, ACS has matured, both in its
feature set as well as in its reliability and performance. However, it is only recently that the ALMA observatory's
hardware and application software has reached a level at which it can exploit and challenge the infrastructure
that ACS provides. In particular, the availability of an Antenna Test Facility(ATF) at the site of the Very Large
Array in New Mexico has enabled us to exercise and test the still evolving end-to-end ALMA software under
realistic conditions. The major focus of ACS, consequently, has shifted from the development of new features
to consideration of how best to use those that already exist. Configuration details which could be neglected
for the purpose of running unit tests or skeletal end-to-end simulations have turned out to be sensitive levers
for achieving satisfactory performance in a real-world environment. Surprising behavior in some open-source
tools has required us to choose between patching code that we did not write or addressing its deficiencies by
implementing workarounds in our own software. We will discuss these and other aspects of our recent experience
at the ATF and in simulation.