The TAO infrastructure can be described with three levels of functionality. At the top level TAO connects the account-based user interface with the user database. In the middle level, the various science modules are then accessed. Both public and private data storage, containing the dark matter simulations, galaxy data, photometry etc., are implemented in the lower level. On the backend TAO is supported by a scalable database cluster hosted on the OzStar (previously gSTAR) supercomputer at the Swinburne University of Technology. Each component of TAO affects the user experience and workflow, speed of mock data generation and retrieval, and the quality and utility of the final mock catalogue.
Users interact with TAO through a simple web form, where they can select a dark matter simulation, galaxy formation model, a box or cone geometry, and the associated parameters which define each. Users select the desired galaxy and simulation properties to be included in the mock, including both absolute and apparent magnitudes in various filters. These selections are then passed to the backend science modules via an XML parameter file, which can also be retrieved from the web interface for reference or later resubmission. Any required computations, e.g. building a light-cone or set of SEDs, are then triggered on the OzStar (previously gSTAR) supercomputer for processing. The user is notified via email when their mock catalogue is completed, which may take from minutes to many hours depending on the size of the task. TAO offers a choice of output formats, including CSV, HDF5 and FITS. The final mock catalogue can then be downloaded directly from the TAO website “History” tab to the user’s local machine. More advanced SQL/ADQL querying of the data is accessible through a VO Table Access Protocol (TAP) client, such as TOPCat.