- Reduces Data Redundancy: Imagine storing a customer’s address every time they place an order. If the address changes, you’d have to update it in multiple places! Normalization helps you store information in one place and refer to it from other places, minimizing repetition.
- Improves Data Integrity: Less redundancy means less chance of inconsistencies. If you update an address in one place, it’s updated everywhere it’s needed.
- Prevents Anomalies: Normalization helps prevent issues like:
- Insertion Anomalies: Difficulty adding new data because you’re missing related information.
- Update Anomalies: Having to update the same information in multiple rows.
- Deletion Anomalies: Accidentally losing valuable information when you delete something seemingly unrelated.
- Easier to Understand and Maintain: A normalized database is generally more logically structured and easier to understand, query, and modify.
Drizzle Relations Fundamentals
In the world of databases, especially relational databases, the concept of relations is absolutely fundamental. Think of “relations” as the connections and links between different pieces of data. Just like in real life, where people have relationships with each other, or objects are related to categories, databases use relations to model how different types of information are connected and work together.
Normalization
Normalization is the process of organizing data in your database to reduce redundancy (duplication) and improve data integrity (accuracy and consistency). Think of it like tidying up a messy filing cabinet. Instead of having all sorts of papers crammed into one folder, you organize them into logical folders and categories to make everything easier to find and manage.
Why is Normalization Important?
Normalization is often described in terms of “normal forms” (1NF, 2NF, 3NF, and beyond). While the details can get quite technical, the core ideas are straightforward:
1NF (First Normal Form): Atomic Values
Goal: Each column should hold a single, indivisible value. No repeating groups of data within a single cell
Example: Instead of having a single address column that stores 123 Main St, City, USA, you’d
break it down into separate columns: street_address, city, state, zip_code.
-- Unnormalized (violates 1NF)
CREATE TABLE Customers_Unnormalized (
customer_id INT PRIMARY KEY,
name VARCHAR(255),
address VARCHAR(255) -- Problem: Multiple pieces of info in one column
);
-- Normalized to 1NF
CREATE TABLE Customers_1NF (
customer_id INT PRIMARY KEY,
name VARCHAR(255),
street_address VARCHAR(255),
city VARCHAR(255),
state VARCHAR(255),
zip_code VARCHAR(10)
);2NF (Second Normal Form): Eliminate Redundant Data Dependent on Part of the Key
Goal: Applies when you have a table with a composite primary key (a primary key made up of two or more columns). 2NF ensures that all non-key attributes are fully dependent on the entire composite primary key, not just part of it.
Imagine we have a table called order_items. This table tracks items within orders, and we use a composite primary key (order_id, product_id)
because a single order can have multiple of the same product (though in this simplified example, let’s assume each product appears only
once per order for clarity, but the composite key logic still applies).
Expand for visual example
CREATE TABLE OrderItems_Unnormalized (
order_id INT,
product_id VARCHAR(10),
product_name VARCHAR(100),
product_price DECIMAL(10, 2),
quantity INT,
order_date DATE,
PRIMARY KEY (order_id, product_id) -- Composite Primary Key
);
INSERT INTO OrderItems_Unnormalized (order_id, product_id, product_name, product_price, quantity, order_date) VALUES
(101, 'A123', 'Laptop', 1200.00, 1, '2023-10-27'),
(101, 'B456', 'Mouse', 25.00, 2, '2023-10-27'),
(102, 'A123', 'Laptop', 1200.00, 1, '2023-10-28'),
(103, 'C789', 'Keyboard', 75.00, 1, '2023-10-29');+------------------------------------------------------------------------------------+
| OrderItems_Unnormalized |
+------------------------------------------------------------------------------------+
| PK (order_id, product_id) | product_name | product_price | quantity | order_date |
+------------------------------------------------------------------------------------+
| 101, A123 | Laptop | 1200.00 | 1 | 2023-10-27 |
| 101, B456 | Mouse | 25.00 | 2 | 2023-10-27 |
| 102, A123 | Laptop | 1200.00 | 1 | 2023-10-28 |
| 103, C789 | Keyboard | 75.00 | 1 | 2023-10-29 |
+------------------------------------------------------------------------------------+Problem: Notice that product_name and product_price are repeated whenever the same product_id appears in different orders.
These attributes are only dependent on product_id, which is part of the composite primary key (order_id, product_id), but not the entire key.
This is a partial dependency.
To achieve 2NF, we need to remove the partially dependent attributes (product_name, product_price) and place them in a separate table where they are
fully dependent on the primary key of that new table.
Normalization to 2NF: Visual explanation
+-------------------+ 1:M +---------------------------+
| Products | <---------- | OrderItems_2NF |
+-------------------+ +---------------------------+
| PK product_id | | PK (order_id, product_id) |
| product_name | | quantity |
| product_price | | order_date |
+-------------------+ | FK product_id |
+---------------------------+CREATE TABLE Products (
product_id VARCHAR(10) PRIMARY KEY,
product_name VARCHAR(100),
product_price DECIMAL(10, 2)
);
CREATE TABLE OrderItems_2NF (
order_id INT,
product_id VARCHAR(10),
quantity INT,
order_date DATE,
PRIMARY KEY (order_id, product_id), -- Composite Primary Key remains
FOREIGN KEY (product_id) REFERENCES Products(product_id) -- Foreign Key to Products
);
-- Insert data into Products
INSERT INTO Products (product_id, product_name, product_price) VALUES
('A123', 'Laptop', 1200.00),
('B456', 'Mouse', 25.00),
('C789', 'Keyboard', 75.00);
-- Insert data into OrderItems_2NF (referencing Products)
INSERT INTO OrderItems_2NF (order_id, product_id, quantity, order_date) VALUES
(101, 'A123', 1, '2023-10-27'),
(101, 'B456', 2, '2023-10-27'),
(102, 'A123', 1, '2023-10-28'),
(103, 'C789', 1, '2023-10-29');3NF (Third Normal Form): Eliminate Redundant Data Dependent on Non-Key Attributes
Goal: Remove data that is dependent on other non-key attributes. This is about eliminating transitive dependencies.
Problem: Let’s say we have a suppliers table. We store supplier information, including their zip_code, city, and state. supplier_id is the primary key.
CREATE TABLE suppliers (
supplier_id VARCHAR(10) PRIMARY KEY,
supplier_name VARCHAR(255),
zip_code VARCHAR(10),
city VARCHAR(100),
state VARCHAR(50)
);
INSERT INTO suppliers (supplier_id, supplier_name, zip_code, city, state) VALUES
('S1', 'Acme Corp', '12345', 'Anytown', 'NY'),
('S2', 'Beta Inc', '67890', 'Otherville', 'CA'),
('S3', 'Gamma Ltd', '12345', 'Anytown', 'NY');+---------------------------------------------------------------+
| suppliers |
+---------------------------------------------------------------+
| PK supplier_id | supplier_name | zip_code | city | state |
+---------------------------------------------------------------+
| S1 | Acme Corp | 12345 | Anytown | NY |
| S2 | Beta Inc | 67890 | Otherville | CA |
| S3 | Gamma Ltd | 12345 | Anytown | NY |
+---------------------------------------------------------------+Solution: To achieve 3NF, we remove the attributes dependent on the non-key attribute (city, state dependent on zip_code) and put
them into a separate table keyed by the non-key attribute itself (zip_code).
Normalization to 3NF: Visual explanation
+-------------------+ 1:M +--------------------+
| zip_codes | <---------- | suppliers |
+-------------------+ +--------------------+
| PK zip_code | | PK supplier_id |
| city | | supplier_name |
| state | | FK zip_code |
+-------------------+ +--------------------+CREATE TABLE zip_codes (
zip_code VARCHAR(10) PRIMARY KEY,
city VARCHAR(100),
state VARCHAR(50)
);
CREATE TABLE suppliers (
supplier_id VARCHAR(10) PRIMARY KEY,
supplier_name VARCHAR(255),
zip_code VARCHAR(10), -- Foreign Key to zip_codes
FOREIGN KEY (zip_code) REFERENCES zip_codes(zip_code)
);
-- Insert data into zip_codes
INSERT INTO zip_codes (zip_code, city, state) VALUES
('12345', 'Anytown', 'NY'),
('67890', 'Otherville', 'CA');
-- Insert data into suppliers (referencing zip_codes)
INSERT INTO suppliers (supplier_id, supplier_name, zip_code) VALUES
('S1', 'Acme Corp', '12345'),
('S2', 'Beta Inc', '67890'),
('S3', 'Gamma Ltd', '12345');Good to know
There are additional normal forms, such as 4NF, 5NF, 6NF, EKNF, ETNF, and DKNF. We won’t cover these here, but we will create a
dedicated set of tutorials for them in our guides and tutorials section.
Database Relationships
One-to-One
In a one-to-one relationship, each record in table A is related to at most one record in table B, and each record in table B is
related to at most one record in table A. It’s a very direct, exclusive pairing.
Use Cases & Examples
- User Profiles and User Account Details: Think of a website. Each user account (in a Users table) might have exactly one user profile (in a UserProfiles table) containing more detailed information.
- Employees and Parking Spaces: An Employees table and a ParkingSpaces table. Each employee might be assigned at most one parking space, and each parking space is assigned to at most one employee.
- Splitting Tables for Organization: Sometimes, you might split a very wide table into two for better organization or security reasons, maintaining a 1-1 relationship between them.
Table A (One Side) Table B (One Side)
+---------+ +---------+
| PK (A) | <---------> | FK (A) | (Foreign Key referencing Table A)
| ... | | ... |
+---------+ +---------+One-to-Many
In a one-to-many relationship, one record in table A can be related to many records in table B, but each
record in table B is related to at most one record in table A. Think of it as a “parent-child” relationship.
Use Cases & Examples
- Customers and Orders: One customer can place many orders, but each order belongs to only one customer.
- Authors and Books: One author can write many books, but (let’s simplify for now and say) each book is written by one primary author.
- Departments and Employees: One department can have many employees, but each employee belongs to only one department.
Table A (One Side) Table B (Many Side)
+---------+ +---------+
| PK (A) | ----------> | FK (A) | (Foreign Key referencing Table A)
| ... | | ... |
+---------+ +---------+
(One) (Many)Many-to-Many
In a many-to-many relationship, one record in table A can be related to many records in table B, and one
record in table B can be related to many records in table A. It’s a more complex, bidirectional relationship.
Use Cases & Examples
- Students and Courses: One student can enroll in many courses, and one course can have many students enrolled.
- Products and Categories: One product can belong to multiple categories (e.g., a “T-shirt” can be in “Clothing” and “Summer Wear” categories), and one category can contain many products.
- Authors and Books: A book can be written by multiple authors, and an author can write multiple books.
Table A (Many Side) Junction Table Table B (Many Side)
+---------+ +-------------+ +---------+
| PK (A) | -------->| FK (A) | <----| FK (B) |
| ... | | FK (B) | | ... |
+---------+ +-------------+ +---------+
(Many) (Junction) (Many)Many-to-many relationships are not directly implemented with foreign keys between the two main tables.
Instead, you need a junction table (also called an associative table or bridging table).
This table acts as an intermediary to link records from both tables.
-- Table for Students (Many side)
CREATE TABLE students (
iid INT PRIMARY KEY,
name VARCHAR(255)
);
-- Table for Courses (Many side)
CREATE TABLE courses (
id INT PRIMARY KEY,
name VARCHAR(255),
credits INT
);
-- Junction Table: Enrollments (Connects Students and Courses - M-M relationship)
CREATE TABLE enrollments (
id INT PRIMARY KEY AUTO_INCREMENT, -- Optional, but good practice for junction tables
student_id INT,
course_id INT,
enrollment_date DATE,
-- Composite Foreign Keys (often part of a composite primary key or unique constraint)
FOREIGN KEY (student_id) REFERENCES students(id),
FOREIGN KEY (course_id) REFERENCES courses(id),
UNIQUE KEY (student_id, course_id) -- Prevent duplicate enrollments for the same student and course
);Why Foreign Keys?
You might think of foreign key constraints as simply a way to validate data - ensuring that when you enter a value in a foreign key column, that value actually exists in the primary key column of another table. And you’d be partially right! This value checking is the mechanism foreign keys use.
But it’s crucial to understand that this validation is not the end goal, it’s the means to a much larger purpose. Foreign key constraints are fundamentally about:
1. Explicitly Defining and Enforcing Relationships
We’ve discussed relationships like One-to-Many between Customers and Orders.
A foreign key is the SQL language’s way of telling the database:
Hey database, I want to enforce a 1-M relationship here. Every value in the customer_id column of the Orders table must correspond to a valid customer_id in the Customers table.
It’s not just a suggestion; it’s a constraint the database actively enforces. The database becomes relationship-aware because of the foreign key.
2. Maintaining Referential Integrity
- This is the core of “data integrity” in the context of relationships. Referential integrity means that relationships between tables remain consistent and valid over time.
- Foreign keys prevent orphaned records. What’s an orphaned record? In our Customer-Order example, an order that exists in the Orders table but doesn’t have a corresponding customer in the Customers table would be an orphan. Foreign keys prevent this from happening (or control what happens if you try to delete a customer with orders - via CASCADE, SET NULL, etc.).
- Why is preventing orphans important? Orphaned records break the logical structure of your data. If you have an order without a customer, you lose crucial context. Queries become unreliable, reports become inaccurate, and your application’s logic can break down.
Example:
Without a foreign key, you could accidentally delete a customer from the Customers
table while their orders still exist in the Orders table. Suddenly, you have orders that point to
a customer that no longer exists! A foreign key constraint prevents this data inconsistency.3. Facilitating Database Design and Understanding
- Foreign keys are not just about technical enforcement; they are also a crucial part of database design documentation.
- When you see a foreign key in a database schema, it immediately tells you:
Table 'X' is related to Table 'Y' in this way.It’s a clear visual and structural indicator of relationships. - This makes databases easier to understand, maintain, and evolve over time. New developers can quickly grasp how different parts of the database are connected.
In essence, foreign key constraints are not just about checking values; they are about:
- Defining the rules of your data relationships
- Actively enforcing those rules at the database level
- Guaranteeing data integrity and consistency within those relationships
- Making your database more robust, reliable, and understandable
Why NOT Foreign Keys?
While highly beneficial, there are some scenarios where you might reconsider or use Foreign Keys with caution. These are typically edge cases and often involve trade-offs.
1. Performance Overhead in Very High-Write Environments
- Scenario: Extremely high-volume transactional systems (e.g., real-time logging, very high-frequency trading platforms, massive IoT data ingestion).
- Explanation: Every time you insert or update data in a table with a foreign key, the database system needs to perform checks to ensure referential integrity. In extremely high-write scenarios, these checks can introduce a small but potentially noticeable performance overhead.
2. Distributed Database Systems and Cross-Node Foreign Keys:
- Scenario: Systems where data is distributed across multiple database nodes or clusters (common in sharded databases, cloud environments, and microservices).
- Explanation: Cross-node foreign keys can introduce significant complexity and performance overhead. Validating referential integrity requires communication between nodes, leading to increased latency. Distributed transactions needed to maintain consistency are also more complex and can be less performant than local transactions. In such architectures, application-level data integrity checks or eventual consistency models might be considered alternatives.
3. Legacy Systems and Data Integration with Non-Relational Data:
- Scenario: Integrating a relational database with older legacy systems or non-relational data stores (e.g., NoSQL, flat files, external APIs).
- Explanation: Legacy systems or non-relational data might not consistently adhere to the referential integrity rules enforced by foreign keys. Imposing foreign keys in such scenarios can lead to data import issues, data inconsistencies, and might necessitate complex data transformation or application-level integrity management instead. You might need to carefully evaluate the data quality and consistency of the external sources and potentially rely on application logic or ETL processes to ensure data integrity instead of strictly enforcing foreign keys at the database level.
You can also check out some great explanations from the PlanetScale team in their article
Polymorphic Relations
Polymorphic relationships are a more advanced concept that allows a single relationship to point to different types of entities or tables. It’s about creating more flexible and adaptable relationships when you have different kinds of data that share some commonality.
Imagine you have an activities log. An activity could be a comment a like or a share.
Each of these activity types has different details. Instead of creating separate tables and
relationships for each activity type and the things they relate to, you might use a polymorphic approach.
Common Scenarios & Examples
- Comments/Reviews: A “Comment” might be related to different types of content: articles, products, videos, etc. Instead of having separate article_id, product_id, video_id columns in a Comments table, you can use a polymorphic relationship.
+---------------------+
| **Comments** |
+---------------------+
| PK comment_id |
| commentable_type | ------> [Polymorphic Relationship]
| commentable_id | -------->
| user_id |
| comment_text |
| ... |
+---------------------+
^
|
+---------------------+ +---------------------+ +---------------------+
| **Articles** | | **Products** | | **Videos** |
+---------------------+ +---------------------+ +---------------------+
| PK article_id | | PK product_id | | PK video_id |
| ... | | ... | | ... |
+---------------------+ +---------------------+ +---------------------+- Notifications: A notification could be related to a user, an order, a system event, etc.
+----------------------+
| **Notifications** |
+----------------------+
| PK notification_id |
| notifiable_type | ------> [Polymorphic Relationship]
| notifiable_id | -------->
| user_id |
| message |
| ... |
+----------------------+
^
|
+---------------------+ +---------------------+ +-----------------------+
| **Users** | | **Orders** | | **System Events** |
+---------------------+ +---------------------+ +-----------------------+
| PK user_id | | PK order_id | | PK event_id |
| ... | | ... | | ... |
+---------------------+ +---------------------+ +-----------------------+Polymorphic relationships are more complex and are often handled at the application level or using more advanced database features (depending on the specific database system). Standard SQL doesn’t have direct, built-in support for enforcing polymorphic foreign key constraints in the same way as regular foreign keys.