dromara/hmily

Javadromara.org

Distributed transaction solutions

tccspringclouddubbomotanspringtansactiontacxa

主要指標

スター成長

スター

4.2k

フォーク

1.4k

週間成長

—

Issue

2k4k

2017年9月2019年2月2020年8月2022年2月2023年8月2025年1月2026年7月

成果物Mavengit clone https://github.com/dromara/hmily.git

README

Financial-level flexible distributed transaction solution

https://dromara.org/

English | 简体中文

Panorama of distributed transaction solutions

Features

high reliability ：supports abnormal transaction rollback in distributed scenarios, and abnormal recovery over time to prevent transaction suspension
usability ：provide zero-invasive Spring-Boot, Spring-Namespace to quickly integrate with business systems
high performance ：decentralized design, fully integrated with business systems, naturally supporting cluster deployment
observability ：metrics performance monitoring of multiple indicators, and admin management background UI display
various RPC ： support Dubbo, SpringCloud, Motan, Sofa-rpc, brpc, tars and other well-known RPC frameworks
log storage ： support mysql, oracle, mongodb, redis, zookeeper etc.
complex scene ： support RPC nested call transaction

Necessary premise

must use JDK8+
TCC mode must use a RPC framework, such as: Dubbo, SpringCloud, Montan

TCC mode

when using the TCC mode, users provide three methods: try, confirm, and cancel according to their business needs. And the confirm and cancel methods are implemented by themselves, and the framework is only responsible for calling them to achieve transaction consistency。

TAC mode

When the user uses the TAC mode, the user must use a relational database for business operations, and the framework will automatically generate a rollback SQL, When the business is abnormal, the rollback SQL will be executed to achieve transaction consistency。

Documentation

If you want to use it, you can refer to Quick Start

About Hmily

Hmily is a flexible distributed transaction solution that provides TCC and TAC modes。

It can be easily integrated by business with zero intrusion and rapid integration。

In terms of performance, log storage is asynchronous (optional) and asynchronous execution is used, without loss of business methods。

It was previously developed by me personally. At present, I have restarted at JD Digital. The future will be a distributed transaction solution for financial scenarios.。