Concurrency control and recovery in database systems
Concurrency control and recovery in database systems
APPC: introduction to LU6.2
Transaction Processing: Concepts and Techniques
Transaction Processing: Concepts and Techniques
Superdatabases for Composition of Heterogeneous Databases
Proceedings of the Fourth International Conference on Data Engineering
Notes on Data Base Operating Systems
Operating Systems, An Advanced Course
Open nested serializability in multidatabase systems
CASCON '92 Proceedings of the 1992 conference of the Centre for Advanced Studies on Collaborative research - Volume 2
Architecture for transaction management in the CORDS multidatabase service
CASCON '93 Proceedings of the 1993 conference of the Centre for Advanced Studies on Collaborative research: distributed computing - Volume 2
Composite multidatabase system concurrency control and recovery
CASCON '93 Proceedings of the 1993 conference of the Centre for Advanced Studies on Collaborative research: distributed computing - Volume 2
Transaction scheduling in dynamic composite multidatabase systems
CASCON '95 Proceedings of the 1995 conference of the Centre for Advanced Studies on Collaborative research
| Hi-index | 0.00 |
Most concurrency control schemes for guaranteeing global serializability in composite multidatabase systems are susceptible to rollbacks. Conservative schemes generate rollbacks because of transaction timeouts, while those of optimistic schemes are caused by certification failures. Typically, rollbacks on any branch of a flat distributed transaction cause a global abort. Global aborts during multidatabase composition degrade performance because of a waste of resources and reductions in multidatabase and local transaction throughputs at component database sites.This paper proposes a failure isolation scheme, based on nested transaction semantics, that attempts to contain multidatabase cell failures to a single cell. Also, given that certain conditions hold, failed transactions may be recovered within a cell and resubmitted without violating global serializability. This recovery scheme is applied to a composite multidatabase prototype.