A database is a program or library that helps you store data. They usually impose some sort of structure to the data to assist with querying and filtering. The most common structure databases use to store data is a table, and most databases will use multiple tables to store data.
A table is composed of rows representing an item, and columns represent some attribute of that item. Much like using a spreadsheet for a task list, where a row would be a task, and you may have a column for a task name, and a second column to determine whether it has been completed.
A relational database is a database in which data is organized into tables, usually with some form of relationship between them. For example, a table containing customer names and addresses, and a table containing sales transactions. Rather than repeat customer details every time a transaction is done, the data is stored once in one table, and then a unique reference is stored in the sales transactions table, linking each transaction to the customer. This approach makes a relational database very efficient for storing data, since reused data can be stored in a single table and referenced in other tables.
Data can be retrieved in many forms by combining multiple tables into new tables using operations like JOIN.
The composed table will have new rows that are a combination of its parent tables.
We use databases to store data. They help provide us guarantees that data is stored and that it exists in the correct format. In addition most databases are heavily optimized making data retrieval very fast.
Because of these attributes, we use databases a lot. Most e-commerce sites use databases to keep inventory and sales records. Doctors offices use databases to store medical records, and the DMV uses databases to keep track of cars. Lawyers use databases to keep track of case law, and many websites use databases to store and organize content. Databases are everywhere, and you interact with them daily.
SQL or Structured Query Language is a :term:`domain specific language` used for accessing databases. It has a declarative syntax, meaning you declare the structure you want returned, the sources, and constraints in a single statement. The database's query parser and planner will determine a how to retrieve your data from this statement.
Many relational databases provide an interactive :term:`CLI` for interacting with the database.
For example, MySQL provides the mysql command, Postgresql provides psql, and Oracle provides sqlplus.
These programs give you the ability to compose queries, and diagnose issues with the software.
todo:: Add example of connecting with each command.
To connect to a MySQL database from the CLI your command would usually take this form:
$ mysql -u username -p -h server-address
Password:Of these flags, -u = username, -p = password, and -h = hostname.
You may provide the password on the command prompt: -ppassword (no space after flag!).
We strongly advise against this, as this may leave the password visible in your shell history.
Most database platforms allow you to create a new database using the CREATE DATABASE SQL query.
It can be executed via a connection to the server, or via the SQL shell.
A MySQL example:
mysql> CREATE DATABASE example_database;Some database platforms have specialized :term:`GUI` tools, or CLI tools.
For example Postgresql has a UNIX command createdb that creates databases.
In many cases these commands are just wrappers around the standard SQL statements.
$ mysqladmin create example_database$ createdb example_databaseSome platforms, like MySQL support multiple databases per instance, while other platforms like Oracle support one database per instance. You should check the documentation for your particular vendor to see what is supported.
Users can be created in most databases using the the CREATE USER statement.
mysql> CREATE USER username;Some relational databases provide additional ways of creating users like specialized command line programs.
MySQL does not support creation of users via the mysqladmin command.
createuser usernamePrivileges can be granted using the SQL GRANT statement. These statements are persisted by the RDBMS when issued. The typical command format is:
GRANT [PRIVILEGE] on [OBJECT] to [USER];The standard SQL privileges are:
| Privilege | Description |
|---|---|
| ALL | Allows user all privileges |
| ALTER | Allows user to alter schema objects |
| CREATE | Allows user to create schema object like tables |
| DELETE | Allows user to delete from an object |
| EXECUTE | Allows user to execute a store procedure or function |
| INSERT | Allows user to add new data to an object |
| REFERENCES | Allows user to create a referential table constraint |
| SELECT | Allows user to read from an object |
| TRIGGER | Allows user to create a trigger |
| TEMPORARY | Allows user to create temporary tables |
| UPDATE | Allows user to update existing data in an object |
Below is an example granting a user SELECT privileges on a table
GRANT SELECT ON TABLE example_table TO user;You can also grant multiple privileges in a single statement.
GRANT SELECT,INSERT ON TABLE example_table TO user;Many databases stray from the SQL standard here, and it is important to read your database's documentation when granting privileges. There may be additional privileges not listed here and syntax can very significantly.
Privileges are removed with the SQL REVOKE statement.
It follows a similar command format like grant:
REVOKE [PRIVILEGE] on [OBJECT] FROM [USER];A normalized schema is a database with a table and column structure designed to reduce data redundancy. Typically data is placed into tables with a unique identifier, or primary key, and then is referenced by id in any tables that wish to use that data.
Suppose we have two types of records in a database; one for a city's population and one for a city's average temperature. We could simply create the tables like so:
City Population:
| City | Population |
|---|---|
| San Francisco | 812,826 |
City Temperature:
| City | Temperature |
|---|---|
| San Francisco | 73 F |
A normalized version of this would have three tables instead of two.
City ID and Name:
| City_id | Name |
|---|---|
| 1 | San Francisco |
City ID and Temperature:
| City_id | Temperature |
|---|---|
| 1 | 73 F |
City ID and Population:
| City_id | Population |
|---|---|
| 1 | 812,826 |
The advantage of this design is that it prevents you from having to enter data multiple times, and generally reduces the storage cost of a row.
If San Francisco changed its name you would only need to update a single row instead of two tables like the first example.
SQL allows you to replace the id with a name using a JOIN statement
when the data is retrieved, making it functionally identical to the two table example.
The SELECT statement is the standard way you read from a table in an SQL database.
You can use it to retrieve a set of data, and perform aggregations on them.
The standard syntax is:
SELECT [column1,column2|*] FROM [TABLE];By adding a WHERE statement, you can have the database filter results:
SELECT user_id, user_name FROM users WHERE user_id = 1;You can join tables using a JOIN statement.
In this example we're temporarily assigning an alias of 'u' for the table users and an alias of 'a' for the table addresses:
SELECT user_id, user_name FROM users u JOIN addresses a
ON u.user_id = a.user_id WHERE user_id = 1;Count the rows in a table by using an aggregation:
SELECT COUNT(1) FROM users;Irder by a column:
SELECT * FROM users ORDER BY user_name;The INSERT statement is used to add additional data into a table.
It can be used to insert data either a row at a time or in bulk.
The standard syntax is:
INSERT INTO table (column1, column2, column3) VALUES (value1, value2, value2)The column list is optional, if you don’t specify which columns you’re inserting data into, you must provide data for all columns.
For example, to insert a single row:
INSERT INTO users (user_name,user_phone) VALUES ("Joe Bloggs","555-1234");Or in bulk:
INSERT INTO users (user_name,user_phone)
VALUES ("John Smith","555-5555"),("Tom Jones","555-0987");Inserting in bulk like that is typically much quicker than using separate queries as the query planner only has to execute once, and any indexes are updated at the end.
UPDATE is the SQL statement for updating existing data in a table.
It should almost always be used with a conditional statement.
The standard syntax is:
UPDATE [TABLE] SET [COLUMN] = {expression}, {COLUMN2={expression}, ...}
[WHERE condition]
[ORDER BY ...]
[LIMIT count];Without a WHERE condition, the statement will apply to all the rows in a table.
Here is a simple example of an UPDATE statement:
UPDATE users SET user_name = 'Jim Smith' WHERE user_name = 'James Smith';DELETE is the SQL statement for removing rows from a table.
The standard syntax is:
DELETE FROM [TABLE]
[WHERE condition]
[ORDER BY ...]
[LIMIT count] ;Note
Without a WHERE condition the statement will apply to all the rows of a table.
Here is a simple example of a DELETE statement:
DELETE FROM users WHERE user_name = 'James Smith';Before doing a write query, run it as a read query first to make sure you are retrieveing exactly what you want. If your query is:
UPDATE users SET disabled=1 WHERE id=1;Run this first to validate you will be affecting the proper record:
SELECT disabled FROM users WHERE id=1;use a
LIMITonUPDATEandDELETE FROMqueries to limit damage imposed by an erroneous queryUPDATE users SET disabled=1 WHERE id=1 LIMIT 1;DELETE FROM users WHERE id=1 LIMIT 1;If your database supports transactions, run
START TRANSACTIONfirst then run your query and check what it has done. If you're happy with what you see then runCOMMITand finallySTOP TRANSACTION. If you realize you've made a mistake, you can runROLLBACKand any changes you've made will be undone.