Support for Fault-Tolerant Query Execution in Distributed Databases

SECONDO is a database management system platform that is available under an open source license. With the help of SECONDO's modular extension concept, new functionality can easily be integrated into the framework. In order to address the scalability requirements that result from the demand to efficiently process huge amounts of data, SECONDO can also be set up as a distributed system across several independent worker nodes, allowing to leverage parallel data processing. However, there is currently no dedicated fault tolerance concept in place. The execution of a distributed query is considered as unsuccessful as soon as one of the involved worker nodes becomes unavailable. In this case, the whole query execution needs to be repeated instead of only repeating the part that has failed, which would be preferable. In the present work, we address the outlined issue by providing the design and implementation of the DBService, a service that supports fault-tolerant query execution in SECONDO. The DBService replicates relations and provides the replicas on request so that the execution of a query is no longer restricted to the original location of the relation.