Method and device for realizing suspension prevention of distributed transactions

Abstract

The invention provides a method for realizing suspension prevention of distributed transactions. The method comprises the following steps of: receiving a first-stage request which comprises a global identifier; if a record which has a rollback state and has the global identifier in the first-stage request exists in a suspension prevention control table, considering that first-stage processing fails; otherwise, enabling a record which has an initial state and has the global identifier in the first-stage request to exist in the suspension prevention control table, and carrying out first-stage business processing; receiving a second-stage rollback request which comprises a global identifier; and enabling a record which has a rollback state and has the global identifier in the second-stage rollback request to exist in the suspension prevention control table, and carrying out business processing of second-stage rollback on the basis of the global identifier in the second-stage rollback request. Through the method provided by the invention, the initiators and participators of distributed transactions can avoid the transaction suspension caused by network jittering and the like, so as to ensure the final consistency of the distributed transactions; meanwhile, under a multi-branch transaction scene, the initiators only need to initiate submission or rollback requests to the participators for once, so that the performance of the distributed transactions is improved.

Description

technical field [0001] The present application relates to the technical field of data management, and in particular to a method and device for realizing anti-suspension of distributed transactions. Background technique [0002] A distributed transaction refers to a transaction involving multiple database operations. Multiple data operations can be operations on different databases, or multiple operations on one database. When a distributed transaction ends, the consistency of the transaction requires that all server operations involved are either all committed or all aborted. In order to achieve this, distributed system transaction processing usually adopts a two-phase commit method. In the first phase, the initiator (or coordinator) sends a request to the participant to ask whether the participant can complete the distributed transaction that he is responsible for. In the transaction branch, the participant replies to the initiator with the result of the success or failure...

AI Technical Summary

Problems solved by technology

If a participant is responsible for multiple distributed transaction branches, the initiator needs to send a two-phase rollback request to the participant for each distributed transaction branch, s

Images