Project presentation

Project objectives

The major objective of DeDiSys is a concept for balancing dependability in distributed software and service systems by partially or temporarily relaxing consistency (data integrity). This concept comprises

• an architecture,
• rules for technology integration including interaction and composition standards for component and service integration,
• the definition of required middleware services, and
• well-defined metrics and evaluation methods for such a system proven by prototype implementations.

By promoting the idea of open systems, we do not aim at a new type of middleware, but rather at the integration of our approach with existing component and service infrastructures and COTS (commercial off the shelf) products. Furthermore, DeDiSys is neither a new method for software testing nor for software verification, but rather a highly innovative method to deal with failures of nodes and links in distributed systems.

Expected impact

Both software engineers and the society in general can benefit from the results of DeDiSys.

The overall framework (‘DeDiSys package’), to be delivered to software engineers (developers, integrators, etc.), comprises a system architecture for each scenario (tightly and loosely coupled), adaptive and hybrid replication protocols, required platform services to build upon, and methods for explicit run-time configuration and negotiation of the consistency/availability trade-off. This framework has the potential to significantly ease building complex dependable distributed systems in combination with well-established or innovative software engineering techniques such as aspect-oriented programming (AOP), model driven architecture (MDA), component-based software engineering (CBSE), and service engineering. Future work will have to aim at reducing the complexity of the reconciliation phase of the tightly-coupled scenario, e.g., with modern software engineering techniques in combination with domain specific languages and platform models.

The society in general will benefit from increased dependability of applications in the following areas:

• Ubiquitous and pervasive computing
• Air traffic control and public safety
• Health care
• Fleet management systems for road traffic, trucks, emergency vehicles
• Control systems in experimental physics
• Security systems
• Grid computing
• Scientific computing