Distributed Transaction Processing Method and Distributed System

Abstract

Embodiments of the present invention provide a distributed transaction processing method and a distributed system. The distributed transaction processing method includes: the coordination subsystem sends a transaction identification application to the management subsystem to obtain the transaction identification, and sends an execution request with the target transaction identification to the execution subsystem; the execution subsystem receives After the execution request, modify the data corresponding to the execution request, and cache the modified data to obtain cached data; the execution subsystem returns the execution result to the coordination subsystem; the coordination subsystem According to the execution result of each execution subsystem, a request for changing the transaction state corresponding to the target transaction identifier is sent to the management subsystem, so as to change the transaction state to a result state; The system acquires the transaction state, and performs a commit or rollback operation on the cached data corresponding to the transaction identifier according to the transaction state.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a distributed transaction processing method and a distributed system. Background technique [0002] In order to alleviate the performance pressure of the stand-alone system and reduce the system size, an effective way is to split the system into multiple subsystems. The large system composed of these subsystems is a form of distributed system. The purpose of distributed transactions is that the execution results of multiple subsystems can succeed or fail in a business operation of a large system. The most basic solution to distributed transactions is the "two-phase commit protocol". The "two-phase commit protocol" has two phases: the "prepare" phase and the "commit / rollback" phase. The "commit / rollback" phase is executed at the end of the entire transaction. Its disadvantage is that if the system is forced to shut down due to downtime or restarting at this stage, the ...

AI Technical Summary

Problems solved by technology

The "commit / rollback" phase is executed at the end of the entire transaction. Its disadvantage is that if the system is forced to shut down due to downtime or restarting at this stage, the final effect of the scheme cannot be achieved, and it is not executed. In the "commit / rollback" phase, the modified data is not confirmed and may be used by other business operations. If the data is wrong but not "rolled back", it will cause system errors

The reason for the above problem is that the "commit / rollback" phase is active

This proactive approach cannot deal with the impact of forced shutdown of the system due to downtime or restart

Images