Summary of 2020

Posted on 2020-12-25 Edited on 2025-07-29 In Others

Farewell 2020

2020 is so special to the world. The influence of COVID-19 seems far more seriously and deeply than many people‘s estimation.

Even if the successful reaction by China government, the epidemic still cause obvious impact to every single man in this country. My salary raise frozen six months, the recruitment market seems shrink, and my company have to take many tough projects to make more money for survive.

In my feelings, time elapse so fast at 2020, one moment I were a huge musk to buy living materials, one moment it's already summer. One moment we celebrate the new year, one moment another Christmas have passed. I continue anxious about future, and continue use coding/reading to restrain it.

At the end of this year, I may say no matter it's a good year or bad year, the earth is still spinning, people we love are still getting old, and our life is still moving forward.

If use one phrase to make a conclusion, I think it will be ”basically satisfied“.

MySQL(InnoDB) 独特的 Repeatable Read 隔离级别

Posted on 2020-11-30 Edited on 2025-07-29 In DB

MySQL 的事务隔离级别

在 MySQL 8.0 Reference Manual 中描述了 InnoDB 的事务隔离介绍： > InnoDB 提供了 SQL-92 标准中描述的所有 4 种隔离级别的支持，它们是：READ UNCOMMITTED，READ COMMITTED，REPEATABLE READ，和 SERIALIZABLE。InnoDB 的默认隔离级别为 REPEATABLE READ。

Spanner：Google 的全球分布式数据库

Posted on 2020-11-06 Edited on 2025-07-29 In TiDB

本文是对《Spanner: Google’s Globally-Distributed Database》的翻译，原文请见：https://static.googleusercontent.com/media/research.google.com/en//archive/spanner-osdi2012.pdf

摘要

Spanner 是 Google 的可伸缩、多版本、全球分布且同步复制的数据库。它是首个在全球范围分布数据且支持外部一致性分布式事务的系统。本文描述了 Spanner 的结构、特性集、其多项设计决策下蕴含的理论基础以及一个能够暴露时钟不确定性（clock uncertainly）的新颖的时间 API。该 API 及其实现是作为关键角色来支撑外部一致性和其他许多涉及整个 Spanner 的强大特性，如对过往数据的非阻塞读（nonblocking read in the past），无锁的读事务，原子的 schema 变更等。