Fault Tolerance using Standard Reflexive Middleware Mechanisms

M.T. Bennani (Tunisia)

Keywords

Faulttolerance, Reflection, Middleware, Portable Intercep tors

Abstract

Lightweight and adaptable applications, essential for wear able and pervasive computing, are not systematically guar anteed when dealing with middleware based fault tolerance features. Some standards middlewares define heavy solu tions not suitable for mobility. For example, Fault Tolerant CORBA (FT-CORBA) standard defines some services, like the replication manager, the fault notifier, etc. to handle the application replication. This approach is heavyweight, and the fault tolerant designed applications are not adapt able. This paper presents how Portable Request Intercep tors (PIs), which we consider as a reflexive1 mechanism, could supply standard CORBA applications with fault tol erance in a lightweight, transparent, and adaptable way. Two distinct approaches are described: integration and ser vice. In the first approach, fault tolerant mechanisms are PIs handled. Conversely, for the second one, PIs act as a ”glue” between fault tolerant service and CORBA applica tions. Finally, conceptual limits, of these two approaches, and their influence on the fault tolerance deployment are discussed.

Important Links:



Go Back