Method and system for realizing distributed transactions

Abstract

The invention discloses a method and system for realizing a distributed transaction, and the method comprises the steps that a transaction initiator node applies for a globally unique and increasing version ID from a version management center, and the version management center sets the state of the version ID as a starting state; the transaction initiator node sends the version ID to the transaction participant node, and when the transaction participant node processes the data, the version ID is attached while the data is stored; and the transaction initiator node judges whether the latest successful state version ID of the data operated by the transaction participation node is smaller than the version ID held by the current transaction or not after the data processing of the transaction participation node is finished, and if so, the version management center updates the version ID state to be a successful state. According to the invention, the consistency of data between different services in a micro-service architecture is realized.

Description

technical field [0001] The present invention relates to the technical field of microservices, in particular to a method and system for realizing distributed transactions. Background technique [0002] Microservice is a very popular architectural style nowadays. It splits a large and complex software application into a group of small, interconnected microservices. Each service implements specific functions and can be deployed independently. Services are loosely connected. coupling. [0003] In the microservice architecture, an operation is jointly completed by a group of different microservice operations, and it is necessary to ensure that this group of operations either all succeed or all fail, which requires a distributed transaction mechanism to provide protection. Simply put, distributed transactions are to ensure the data consistency of different services. [0004] The existing distributed transaction schemes and their existing problems are as follows: [0005] 1) 2PC...

AI Technical Summary

Problems solved by technology

[0013] The main problem of the 2PC scheme is that during the execution of distributed transactions, all node resources participating in the transaction are locked, and the system does not allow other nodes to access the locked resources. It is difficult to further improve the execution efficiency of the system under such execution ; When submitting in the second step, participants are not allowed to make mistakes. If messages are lost, etc., it will lead to data inconsistency

[0022] In 2PC, only the coordinator sets the timeout (that is, the participant does not reply to the response message within the specified time)

[0024] The main problem of 3PC is: 3PC is an extension and supplement to 2PC, but it does not effectively solve the problem of performance and data inconsistency, and it also increases the complexity of implementation

[0038] The main problem of the Saga scheme is that it does not provide transaction isolation, transactions are interleaved, and partial results of other transactions will be seen, which can be understood as an imperfect compensation transaction mechanism

Images