How different business scenarios database level segmentation?

Original link: https://gitbook.cn/m/mazi/activity/58ecce75b80b10d84f8bf763?isSeries=yes\x26amp;utm_source=news191027_3

Many Internet business, has a large amount of data features, with the gradual increase in the amount of data, the database system has gradually become the bottleneck.

Master-slave synchronization scheme can only read and write separation of infrastructure to enhance the read performance of the database, the expansion of the amount of data in one database, and write performance bottlenecks are not well resolved. At this level of database segmentation techniques born out how horizontal segmentation under different business scenarios, you need to pay attention to the technical point of segmentation process, after slicing encountered new problems and solutions, is the topic to be discussed The problem.

The topic achieve the desired effect: From now, no matter what business scenarios, data volume level segmentation technical problems since then will not have to worry about .

Scan code to understand the topic Chat

▼

Author Shen Jian

Public Author "architect of the Road", 58 home Senior Director, Chairman of the Technical Committee. Engineering before Baidu, 58 city overhead, Chairman of the Technical Committee.

Section describes

Section I: Database typical architecture practice

This is an introduction, the Internet database schema design, the horizontal slicing technology-related concepts and doubts: What is the level of segmentation and vertical segmentation, what is the grouping, what is fragmentation, what is the routing rules, what are common routing rules, what their advantages and disadvantages, the level of segmentation methodology, the difficulties often encountered is doing a stage for the subsequent four games.

Covering 90% of Internet traffic characteristics of four types of business, user-centered ( "single-KEY" class of business), center post ( "one to many" business class), friendship ( "many to many" business class), Order Center ( " multi KEY "type service) how to implement each level of segmentation.

Section II: starting from the User Center, talk "single-KEY" class business database level segmentation architecture practice

There is a class of service features "single KEY", a typical representative of the "Center" this business scenario, the user with the increasing amount of data, significant decrease in performance database, how segmentation is horizontal field Chat service center to the user Key:

1. Features scenes with "single-KEY" class business.

2. How to "single-KEY" horizontal segmentation class of business core guiding ideology.

3. “单KEY”类业务水平切分后遇到的潜在问题（最典型的问题->通过userid来切分，username上的查询怎么办？）。

4. “单KEY”类业务水平切分最佳实践。

第三节：从帖子中心开始，聊“1对多”类业务数据库水平切分架构实践

有一类“1对多”特征的业务，典型代表是“帖子中心”，1个帖子对应1个发布用户，1个发布用户能够发布多个帖子，这个业务场景，随着用户数据量越来越大，数据库性能显著降低，如何来对帖子中心业务进行水平切分是本章的重点：

1. “1对多”类业务的特点与场景。

2. “1对多”类业务如何进行水平切分核心指导思想。

3. “1对多”类业务水平切分后遇到的潜在问题（最典型的问题->通过tieziid来切分，userid上的查询怎么办？）。

4. “1对多”类业务水平切分最佳实践。

第四节：从好友关系开始，聊“多对多”类业务数据库水平切分架构实践

有一类“多对多”特征的业务，典型代表是“好友关系”，1个用户能加多个好友，1个用户又能被多个人反加好友，这个业务场景，随着用户数据量越来越大，数据库性能显著降低，如何来对好友关系业务进行水平切分是本章的重点：

1. “多对多”类业务的特点与场景。

2. “多对多”类业务如何进行水平切分核心指导思想。

3. “多对多”类业务水平切分后遇到的潜在问题（最典型的问题->通过userid来切分，friendid上的查询怎么办？）。

4. “多对多”类业务水平切分最佳实践。

第五节：从订单中心开始，聊“多KEY”类业务数据库水平切分架构实践

有一类“多KEY”特征的业务，典型代表是“订单中心”，业务查询维度会覆盖order_id/buyer_id/seller_id，这个业务场景，随着用户数据量越来越大，数据库性能显著降低，如何来对订单中心业务进行水平切分是本章的重点：

1. “多KEY”类业务的特点与场景。

2. “多KEY”类业务如何进行水平切分核心指导思想。

3. “多KEY”类业务水平切分后遇到的潜在问题（最典型的问题->通过order_id来切分，buyer_id/seller_id上的查询怎么办？）。

4. "Multi-KEY" level segmentation class of business best practices.

Section VI: after the words: In addition to the level of segmentation, database schema design often what problems

This is a summary, in the Internet database schema design, in addition to experiencing the level of cut classification, availability will encounter many problems, read performance, consistency, SQL, etc., what ideas to solve these problems is that the text decomposition Listen .

What can you get?

Record Collection in Shen Jian have done the following questions answered

• There is a kind of typical operation is a list of page operation, this process like? What is the impact on the sub-library sub-table strategy?

• Metadata is what is meant is the basis of data from a database? Or refer to the Notes programming model?

• To database fragmentation preparatory work what it is, how to distinguish between the tables to be fragmented?

• The article refers to what libraries are for different database software?

• How distributed transaction processing?

• Sub-library sub-table with how to use the service in the micro mode?

• Similar user table, authority table, the department table, the service needs to be done to make micro redundancy it?

• The purpose is to make redundant query performance data in order to solve large time?

• Many relationships in addition to redundant data, there are other ways to deal with it?

• Distributed development, many to many data, how to encapsulate API available? Security encryption and decryption of how to do?

• Now 58 home interior is rpc develop their own, or with springcloud + springboot?

• Information on how to desensitization 58 is designed?

• Two programs of integrated programs, can play this particular the specific program do?

• 50 million in a single day can be designed in Log in MySQL do?

• If the online business system has been used in "multi-key" table should be split on how to proceed?

• And the number of multi-paged pool count statistics, how to help each library statistics and inquiry?

• The client library with sub-sub-sub-table and the server have any good library sub-table, how the selection?

• Eventual consistency there any good middleware software? Implement own algorithm takes a lot of time?

Above these issues, not simply a few words can say clearly, to which everyone interested in the topic are welcome to subscribe to this Chat topic.

Click to read the original , subscribe Chat Shen Jian big brother.