Database Learning Notes
This repo contains course notes in the following courses:
- Database Systems Concepts & Design from Georgia Tech on Udacity
- Intro to Relational Database on Udacity
- SQL for Data Science from University of California, Davis on Coursera
Benefits
Why not just fetch all the data from DB to my application code, and do all the data operations there?
-
Speed
DB do the operations much faster.
-
Space
Usually the data scale is too large to fit in memory, or will take up too much memory.
DB Types
Relational DB (关系型数据库) => Relational DB Management System (RDBMS) (关系型数据库管理系统)
- SQLite
- 轻量级、嵌入式DB
- 其一个DB就是一个文件(
xxx.db), 存储在disk上 - => 经常被集成到桌面和移动端应用之中
- 不能承受高concurrency访问
- MySQL
- 为server-side设计的, 使用最广泛的DB server
- 能承受高concurrency访问
- PostgreSQL
- 为server-side设计的, 使用流行度呈上升趋势
- 能承受高concurrency访问
- Oracle
- 不开源、付费
优点:
- data的每个table都非常规整, 确保了每一个row (record) 都有相同数量的column (field)
- 通过foreign key的使用, 可以减少数据冗余
缺点:
- 很多时候为了拿到某个information, 需要把多个table join起来, 这是一个很耗时且占用资源的操作
Not-only SQL (NoSQL)
Key-value store (键-值存储)
适用于要存储的data type相对简单, 但需要极高的retrieve和insert速度的嵌入式场景
- Redis
-
In-memory存储
-
存储的data有结构, 用来存储
String,List(linked-list),Hash(hash table),Set(set),SortedSet(tree set) 等data type -
=> 常用作data structure server
-
由于是in-memory存储, 所以也常常用作在application layer和实际DB (比如MySQL) 之间的一层cache来使用
-> 详情看Redis overview页面
-
Document Store (文档存储)
Basic Concepts:
| Document store | Relational DBMS |
|---|---|
| DB | DB (same) |
| Collection (集合) | Table |
| Document (文档) | Record |
- MongoDB
- 能承受高concurrency访问
- => 多用于data的采集和分散处理 (Map/Reduce), 特别是在大数据处理方面比较擅长
Graph Store (图存储)
e.g., 每个entity对应一个node; 每个relation对应一条edge
License
This repo is distributed under the MIT license.