The applications consist of lots of different components such as aero-elastic simulation work flow. According to Hempel, it is very tedious work. The management of complex simulation applications is a hard problem because the component codes and data are geographically distributed. A single run has many input and output files and interactions. In addition, there is an institution-wide organisation required.

The TENT-DataFinder solution consists of a DM Tool DataFinder to organise data and metadata. The integration system TENT configures, executes and controls complex work flows. The user perspective of TENT provides an environment for numerical simulation, from which the whole operation can be steered. Essential features include easy set-up and configuration of simulation work flows. Rolf Hempel also mentioned the use of distributed computing resources, on-line steering and visualization and project-based data management with support for co-operative work.

The IT perspective of the system has also essential characteristics. It is component-based with a distributed component model based on CORBA, an extensible Java-GUI, Grid technology for distributed computing, a flexible integration of the existing tools, and data management with open solutions including WebDAV and XML which can be integrated with DataFinder.

TENT is an integration system for numerical simulation work flows deployed at the German Aerospace Centre. It is a component system based on CORBA. The system offers techniques for wrapping arbitrary applications. TENT provides a graphical construction of work flows with distributed environments or Grids. TENT helps researchers with on-line visualization and steering to gain insight on the manufacturing of turbine engines, rocket motors, automobile design, virtual aircraft design, space re-entry, and flight manoeuvres.

The data management problem at DLR is that there is no central data management policy. Every employee organises the data individually. Researchers spend about 30% of their time searching for data. There is a problem with data left behind by temporary staff. There is also a rapidly growing volume of simulation and experimental data and there are, of course, the legal requirements for the long term availability of data to up to 50 years, as Rolf Hempel stated.

One example is the Institute of Aerodynamics and Flow Technology which performs numerical and experimental studies in the aerospace sector at three different sites. At the Brunswick site there is an 80% increase in data volume between 2002 and 2003. We are talking 13.3 TB of data here. Therefore, the institute initiated the search for a DLR data management solution.

The requirements for innovative data management are scalability in terms of data volume and the number of files. The support for collaborative work and access control management is also necessary as is data organisation according to the project structure. A flexible annotation system and a search mechanism are needed. Rolf Hempel also mentioned support for on-line and off-line media, and for various client systems. DLR also asked for the automation of simple work flows. The overall objectives were therefore to find the data faster and to reduce the data management cost.

First there was an evaluation of the commercial product data management (PDM) solutions in the market. Two products of major software vendors were analysed. The first system was eliminated after one week trial because of substantial problems with the user interface. The second one ran in a one-week proof of concept test at the institute in which a helicopter rotor was simulated with the block-structured CFD solver FLOWer. Rolf Hempel said that the PDM system solved the problem but it has many unused features and it is very expensive.

The alternative concept from DLR SISTEC is a client-server solution based on open and stable interface standards, such as XML and WebDAV. It uses standard interfaces to allow easy use op open source and commercial software components and solves the problem as good as the PDM system but at a much lower cost.

This DataFinder functionality includes clients for various platforms, free data modelling and versioning, user/role/access management, a notification system and an extensive search mechanism, and is extensible through Python scripts. The DataFinder software consists of existing software with many options available. The interface requirement is supported by the http/WebDAV protocol. It is possible to store files and metadata separately. There are options for metadata storage including an Apache web server and an XML database.

There has been the release of a first stable product version in October 2003. The installation at two DLR institutes in Brunswick and G–ttingen is under way. There are ongoing proof-of-concept projects at several other DLR institutes. Rolf Hempel also unfolded the plans for new DataFinder features such as Web interface and the coupling with full text search engines. In addition, DLR started negotiations with commercial partners for external marketing of the DataFinder.

http://www.dlr.de/sc/produkte/TENT

http://www.dlr.de/DataFinder