Chinese Journal of Computers   Full Text
Title	Fault Tolerance Configuration for Middleware Services
Authors	LI Jun-Guo HUANG Gang ZOU Jian MEI Hong
Address	(Key Laboratory of High Confidence Software Technologies (Peking University) of Ministry of Education, Institute of Software, School of Electronics Engineering and Computer Science, Peking University, Beijing 100871)
Year	2007
Issue	No.10(1696—1704)
Abstract &   Background	Abstract Reliability of middleware significantly impacts that of distributed applications running on it. Software-implemented fault tolerance is regarded as one of the effective means to improve reliability of middleware. But a middleware usually employs only one fault tolerance mechanism for some given scenarios in past work. This paper proposes a runtime software architecture based approach for applying fault tolerance capability for middleware services in a semi-automatic manner. It does following contributions: First, runtime software architecture helps to automatically construct the component dependency view and error propagation view for a middleware based system, and hence, helps developers or maintainers to identify faulty services. Second, developers or maintainers can manipulate runtime software architecture to configure error detection mechanisms and recovery mechanisms. At last, the configured fault tolerance mechanisms are weaved into middleware with aspect oriented programming techniques. The authors provide an Eclipse-based GUI tool to support the approach and demonstrate the tool on a J2EE (Java 2 platform Enterprise Edition) middleware. keywords middleware; reliability; software architecture; software implemented fault tolerance background Middleware has already become a common infrastructure for distributed computing. The reliability of middleware has a crucial impact on that of distributed applications running on it. It is accepted that software-implemented fault tolerance can improve reliability of middleware as shown in past work. However, a middleware can employ only one kind of fault tolerance mechanism in some scenarios in existing work. It is believed that a configurable fault tolerant capability is needed for complex distributed applications. So, the paper proposes a runtime software architecture based approach to configure fault tolerance mechanisms for middleware services in a semi-automatic manner. Specifically, the two views of runtime software architecture, component dependency view and error propagation view, give developers or maintainers a global abstraction for the whole system. Fault tolerance mechanisms are added to middleware using software architectural concepts and operations. In addition, the authors provide a GUI tool on J2EE middleware to support above idea and present a case study. This work is sponsored by the National Basic Research Program(973 Program) of China under grant No.2005CB321805, and the National Natural Science Foundation of China under grant Nos.90612011 and 90412011. The team has made some important progresses on software architecture and J2EE middleware. ABC(Architecture Based Component Composition) approach and PKUAS are two major progresses that formed the foundation of our study. The team also addresses the quality of distributed system.