ERCIM News No.35 - October 1998
by Peter Fankhauser
Enterprises are finding that they need to make better use of their information resources in order to operate effectively in contemporary business or public environments. Therefore, existing islands of information need to be integrated within a flexible and coherent technical framework. One prominent approach towards this end are Intranet and Internet platforms that support access to existing files, databases, data warehouses, and legacy applications from a single browser interface. However, the existing tools do not allow for the integration of multiple data sources beyond simple links between them. MIRO-Web is an Esprit Project (EP 25208) that aims at the development of middleware components providing integrated access to multiple data sources from standard Web browsers (eg Netscape or Microsoft Explorer) using Internet/Intranet protocols.
MIRO-Web builds the approach for database interoperability developed in the ESPRIT Project IRO-DB (EP 8629). With this approach, interoperability is accomplished by means of three layers: Database adapters provide for a uniform access to individual databases, a communication layer supports remote data access, and an interoperable layer allows for integrated views on several databases. Adopting this approach to Internet and Intranet platforms poses two fundamental challenges.
The majority of the information sources are not fully fledged databases with an explicit schema, but semi-structured data of often proprietary format. Especially Internet sources evolve at a much higher pace than databases in a controlled business setting. MIRO-Web puts particular focus on tackling the large variety of sources and their dynamics. Towards this end, the following three components will be developed:
- a data source adapter construction toolkit, providing the means to build adapters for accessing a wide variety of local data sources, including structured or weakly structured files (eg, Cobol or HTML), relational or object-relational databases, and legacy hierarchical databases
- a metadata integrator workbench to extract and integrate various schemas of data sources in terms of object-relational types and tables, and to populate a metadata repository containing data sources and integrated view definitions.
- a mediator to process object-relational queries (ie SQL3), decompose them into sub-queries addressed to data source adapters, and return integrated and structured results as object-relational collections capable of being displayed by a Web browser or manipulated by an application. In parallel a datawarehouse solution based on an object-relational DBMS will be developed, which caches portions of the individual datasources for better performance.
Together these components form a middleware layer that allows for uniform access to widely differing data sources, and also merging multiple data sources into a single integrated view.
Two typical Internet and Intranet applications will be prototyped to better design and commercially assess the technology during the project. The first application for Tyrol Information Systems (TIS GmbH) realizes a touristic information and reservation virtual server which accesses several tourism webservers and reservation systems. The second application for the Basque Health Service (BHS) aims at a pilot system that allows both health professionals of Primary Care and Outpatient Clinics to consult hospital health information from a variety of systems in an integrated fashion.
The project has started in November 1997. It is carried out in collaboration with Ibermática (Spain), Bull (France), OSIS (France), the Basque Health Service (Spain), TIS (Austria), and INRIA (France). Information on the web at : http://www.darmstadt.gmd.de/oasys/projects/miro/mirowe.html
Peter Fankhauser - GMD
Tel: +49 6151 869 940