GGrantIndex
← Search

CAREER: Integrated Fault Tolerance and Real-Time Support for Middleware Applications

$434,800FY2003CSENSF

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

This project exploits the CORBA middleware standard as a vehicle for exploring the composition of real-time and fault tolerance for distributed middleware applications. Although the initial approach will apply to CORBA, the results will be applicable to other middleware, such as Java. The primary scientific contributions of this project are the description of the precise trade-offs that occur when both real-time and fault tolerance support are required for a middleware application, along with an architecture and strategies that can be used to resolve these trade-offs. The primary technical contributions of the project are a new set of practices and algorithms for building real-time fault-tolerant systems, and tools for sanitizing non-determinism in middleware applications, with the aim of meeting both real-time predictability and fault-tolerant consistency requirements. This project furthers our technical understanding of the precise trade-offs and complex interactions between ``-ilities'' such as real-time and fault tolerance, thereby helping to reduce the number of proprietary brittle systems that continue to be built. The project has the potential to influence the state of the art and the state of the practice, enabling future COTS-based mission-critical applications to be developed with simultaneous real-time and fault-tolerant support. Recognizing that applications vary in their dependability needs, this project is unique in that it provides mechanisms to mix-and-match the different fault tolerance mechanisms in a manner that allows for the customization of the system's perceived dependability. Through participation in organizations such as the Object Management Group, the results of this project will be made available to harden ongoing and future standards on fault tolerance and real-time for middleware. Elements of this project will be incorporated into curricula in order to expose students to the cross-cutting concerns of both reliability and real-time. Students who are exposed to not just one ``-ility'', but to the conflicts and trade-offs between multiple ``-ilities'', will make better and more discriminating software/system architects who can be entrusted to build the mission-critical systems of the future.

View original record on NSF Award Search →