Browsing by Author "Ulmala, Minna"

eBusiness collaboration and an eBusiness process are introduced as a context of a long running eBusiness transaction. The nature of the eBusiness collaboration sets requirements for the long running transactions. The ACID properties of the classical database transaction must be relaxed for the eBusiness transaction. Many techniques have been developed to take care of the execution of the long running business transactions such as the classical Saga and a business transaction model (BTM) of the business transaction framework. Those classic techniques cannot adequately take into account the recovery needs of the long running eBusiness transactions and they need to be further improved and developed. The expectations for a new service composition and recovery model are defined and described. The DeltaGrid service composition and recovery model (DGM) and the Constraint rules-based recovery mechanism (CM) are introduced as examples of the new model. The classic models and the new models are compared to each other and it is analysed how the models answer to the expectations. Neither new model uses the unaccustomed classification of atomicity even if the BTM includes the unaccustomed classifying of atomicity. A recovery model of the new models has improved the ability to take into account the data and control dependencies in the backward recovery. The new models present two different kinds of strategies to recover a failed service. The strategy of the CM increases the flexibility and the efficiency compared to the Saga or the BTF. The DGM defines characteristics that the CM does not have: a Delta-Enabled rollback, mechanisms for a pre-commit recoverability and for a post-commit recoverability and extends the concepts of a shallow compensation and a deep compensation. The use of them guarantees that an eBusiness process recovers always in a consistent state which is something the Saga, the BTM and the CM could not proof. The DGM offers also the algorithms of the important mechanisms. ACM Computing Classification System (CCS): C.2.4 [Distributed Systems]: Distributed applications