A Framework for the Design and Implementation of Fault-Tolerant Distributed and Mobile Information Systems

Distributed and Mobile Information Systems are becoming an integral part of almost every large establishment such as Manufacturing Industries, Service Industries, Financial and Educational Institutions and Government organizations. Distributed and mobile database systems serve as back-end for these information systems. Hence, ensuring these databases cope with failures transparently is critical for the successful and efficient operation of these establishments. This project deals with the design and implementation of ef- ficient checkpointing methods for recovering from failures transparently in distributed and mobile database systems as well as distributed object-based systems. The issues involved in the design and implementation of efficient checkpointing and recovery techniques for distributed computations is well understood; efficient checkpointing and recovery algorithms have been developed in this area. However, to our knowledge, the issues involved in designing efficient non-intrusive checkpointing methods for distributed database systems have not been addressed very well so far. Moreover, such issues have not been looked into in the areas of mobile database systems and distributed object-based systems. In this project, we propose to investigate these issues and find answers. Intellectual merit of the proposed activity: The proposed research focuses on developing efficient techniques for achieving fault-tolerance in distributed and mobile database systems, and distributed object-based systems and spreads in the following directions: (i) We propose to develop first a theoretical frameworktha t addresses the issues involved in designing efficient non-intrusive checkpointing techniques for distributed database systems that would facilitate the construction of transaction-consistent checkpoints on-the-fly. We will study the suitability of the developed frameworkf or mobile database systems and refine it if necessary to suit mobile database systems. We will also develop a theoretical frameworkthat addressees the issues involved in designing efficient checkpointing and recovery techniques for distributed object-based systems. (ii) Based on the theoretical framework, we propose to develop efficient non-intrusive checkpointing techniques for distributed database systems, that would facilitate the construction of transaction-consistent checkpoints on-the-fly which in turn would help healing the system from failures transparently. We will also develop efficient checkpointing and recovery techniques for mobile databases and object-based systems. (iii) We will evaluate the performance of the developed checkpointing and recovery techniques through extensive experimental evaluation. (iv) We will disseminate the results of our research by presenting in international conferences and publishing in international journals. (v) We will integrate our research findings in the graduate curriculum. Important expected broader impacts of this project: (i) The outcome of this project will significantly enhance the understanding of the issues involved in designing non-intrusive checkpointing and asynchronous recovery techniques for distributed and mobile database systems, and distributed object-based systems. (ii) It will result in the availability of improved techniques for checkpointing and recovery for distributed and mobile database systems, and distributed object-based systems; the efficiency of the developed techniques will be supported by hard data from experimental evaluation as well as strong theoretical foundation; this in turn will help build reliable distributed and mobile database systems, and distributed object-based systems. (iii) Availability of improved techniques for achieving fault-tolerance in distributed and mobile database systems will help in the successful and efficient operation of various organizations such as Manufacturing Industries, Service Industries, Financial and Educational Institutions and Government Organizations. (iv) Training graduate and undergraduate students in the area of fault-tolerance in distributed and mobile database systems will help them in their career to deal with issues related to fault-tolerance in distributed and mobile database systems and objectbased systems. (v) Integration of the results of our research in a graduate level course would help the students acquire the tools and techniques necessary for the design and implementation of fault-tolerant distributed and mobile database systems and object-based systems.