SQL Tables

Basics

You can easily diagram entity-relationship diagrams (ERDs) in D2 by using the sql_table shape. Here's a minimal example:

1my_table: {
2  shape: sql_table
3  # This is defined using the shorthand syntax for labels discussed in the containers section.
4  # But here it's for the type of a constraint.
5  # The id field becomes a map that looks like {type: int; constraint: primary_key}
6  id: int {constraint: primary_key}
7  last_updated: timestamp with time zone
8}

Each key of a SQL Table shape defines a row. The primary value (the thing after the colon) of each row defines its type.

The constraint value of each row defines its SQL constraint. D2 will recognize and shorten:

constraint	short
primary_key	PK
foreign_key	FK
unique	UNQ

But you can set any constraint you'd like. It just won't be shortened if unrecognized.

info

You can also specify multiple constraints with an array.

1x: int { constraint: [primary_key; unique] }

Foreign Keys

Here's an example of how you'd define a foreign key connection between two tables:

1objects: {
2  shape: sql_table
3  id: int {constraint: primary_key}
4  disk: int {constraint: foreign_key}
5
6  json: jsonb {constraint: unique}
7  last_updated: timestamp with time zone
8}
9
10disks: {
11  shape: sql_table
12  id: int {constraint: primary_key}
13}
14
15objects.disk -> disks.id

info

When rendered with the TALA layout engine or ELK layout engine, connections point to the exact row.

Example

Like all other shapes, you can nest sql_tables into containers and define edges to them from other shapes. Here's an example:

1cloud: {
2  disks: {
3    shape: sql_table
4    id: int {constraint: primary_key}
5  }
6  blocks: {
7    shape: sql_table
8    id: int {constraint: primary_key}
9    disk: int {constraint: foreign_key}
10    blob: blob
11  }
12  blocks.disk -> disks.id
13
14  AWS S3 Vancouver -> disks
15}