Next: 2 Proposed Architecture Up: A LightweightMessage-Oriented Application Previous: A LightweightMessage-Oriented Application
During the last five years, a lot of different database and information systems have been developed with an intention to offer their functionalities to a wide range of users via the Internet and especially via the World Wide Web (WWW). Many more information systems do not even provide a Web interface. Efforts are made in the field of Federated Information Systems (FIS) and by Commercial Application Servers (CAS) to integrate those heterogeneous systems.
FIS accomplish this by a federated model or scheme on top of the systems to be integrated. Our approach differs from the one used in FIS in the way that we do not integrate semantically the backend systems. Instead, we introduce a general data description model which is only responsible for the presentation of data, whereas the semantics is determined by the individual backend system (BES). The data description model is part of a middleware component that encapsulates all communication aspects and therefore can be seen as a means to hide the technical heterogeneity of the BES. Virtually any kind of source system with information retrieval capabilities or database management system can be considered a BES. Our approach should yield a lightweight information system (explained in Section 2.2) with which any BES can be rapidly adapted to our system. After the adaption, each BES remains an autonomous system but additionally offers its functionality to a wide range of Web users through an uniform Web interface. Logical changes to the BES do not affect our middleware because of their autonomity. On the other hand, the missing integrating logical view prevents a direct reuse of data or any combination of the information provided by the BES. We are therefore not concerned with the development of a meta model that describes an integrated view of the incorporated systems. Our ``meta model'' just determines the messages that must be delivered between the client (the uniform Web interface) and the individual BES.
In contrast to CAS, our system intends to offer a facility for forming collections of individual heterogeneous systems with the help of integration mechanisms. A collection (i.e. each individual system within the collection) can be accessed via a single uniform Web interface because the structure of the transferred messages is general and different messages can be mapped onto the same view, like different XML-documents  can be displayed within a single viewer. The communication of most CAS is based on concepts like CORBA  or Java's  Remote Method Invocation (RMI) with compiled stubs on the client and server side which does not provide an uniform access or view because each stub must be handled individually. CAS provide uniform interfaces only for the management of the integrated systems. Similiar to CAS, our system architecture allows to solve communication problems like scalability and bandwidth. Additionally our system solves the problem of uniform access to the individual BES.
All these problems arise in an environment where a large number of users want to interact with a large number of different information servers, which is the typical situation on the Internet and the WWW. All the problems of an uniform interface, of scalability and of bandwidth are explained and discussed, together with the design and implementation of our system, in the next two sections.
Section 2 describes the architecture of our lightweight, message-oriented system. Section 3 explains how the different communication problems mentioned above are solved by the system, and Section 4 shows the integration of a scientific method base system (PROGRESS) and an XML-based monitoring system (SPECTO) as examples of how users can interact with our system. The article concludes with a short summary.
Next: 2 Proposed Architecture Up: A LightweightMessage-Oriented Application Previous: A LightweightMessage-Oriented Application Ralf-Dieter Schimkat
Thu Dec 9 14:08:00 GMT+1 1999