A Transactional Model for Long-Running Activities
Umeshwar Dayal, Meichun Hsu, Rivka Ladin
Abstract
Umeshwar Dayal, Meichun Hsu, Rivka Ladin
Abstract
Many computer-supported applications are of long duration and consist of multiple steps that are exe-cuted over possibly heterogeneous servers. Such activ-ities have weaker atomicity requirements than trans-actions. Previously, we illustrated how to organize the execution of such activities using triggers and transac-tions. In this paper, we describe an execution model in.which activities may consist recursively of steps that may be subactivities or transactions. The model defines precisely the semantics of activities: commu-nication between steps and the failure semantics of activities including compensation and exception han-dling. The model also supports querying the status of activities. We also propose an implementation of the model using recoverable queues for reliably chaining the steps according to the semantics of the model. 1
OpenAlex reports 164 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Many computer-supported applications are of long duration and consist of multiple steps that are exe-cuted over possibly heterogeneous servers. Such activ-ities have weaker atomicity requirements than trans-actions. Previously, we illustrated how to organize the execution of such activities using triggers and transac-tions. In this paper, we describe an execution model in.which activities may consist recursively of steps that may be subactivities or transactions. The model defines precisely the semantics of activities: commu-nication between steps and the failure semantics of activities including compensation and exception han-dling. The model also supports querying the status of activities. We also propose an implementation of the model using recoverable queues for reliably chaining the steps according to the semantics of the model. 1
Key concepts: Atomicity, Computer science, Chaining, Server, Semantics (computer science), Transaction processing, Distributed computing, Transactional leadership