Customize your data model

Amplify automatically creates Amazon DynamoDB database tables for GraphQL types annotated with the @model directive in your GraphQL schema. You can create relations between the data models via the @hasOne, @hasMany, @belongsTo, and @manyToMany directives.

Setup database tables

The following GraphQL schema automatically creates a database table for "Todo". @model will also automatically add an id field as a primary key to the database table. See Configure a primary key to learn how to customize the primary key.

type Todo @model {
  content: String
}

Upon amplify push or cdk deploy, Amplify deploys the Todo database table and a corresponding GraphQL API to perform create, read, update, delete, and list operations.

In addition, @model also adds the helper fields createdAt and updatedAt to your type. The values for those fields are read-only by clients unless explicitly overwritten. See Customize creation and update timestamps to learn more.

Try listing all the todos by executing the following query:

query QueryAllTodos {
  listTodos() {
    todos {
      items {
        id
        content
        createdAt
        updatedAt
      }
    }
  }
}

import { Amplify } from 'aws-amplify';
import { generateClient } from 'aws-amplify/api';
import config from './amplifyconfiguration.json';
import { listTodos } from './graphql/queries';

const client = generateClient();

Amplify.configure(config);

try {
  const result = await client.graphql({ query: listTodos });
  const todos = result.data.listTodos;
} catch (res) {
  const { errors } = res;
  console.error(errors);
}

Configure a primary key

Every GraphQL type with the @model directive will automatically have an id field set as the primary key. You can override this behavior by marking another required field with the @primaryKey directive.

In the example below, todoId is the database's primary key and an id field will no longer be created automatically anymore by the @model directive.

type Todo @model {
  todoId: ID! @primaryKey
  content: String
}

Without any further configuration, you'll only be able to query for a Todo via an exact equality match of its primary key field. In the example above, this is the todoId field.

Note: After a primary key is configured and deployed, you can't change it without deleting and recreating your database table.

You can also specify "sort keys" to use a combination of different fields as a primary key. This also allows you to apply more advanced sorting and filtering conditions on the specified "sort key fields".

type Inventory @model {
  productID: ID! @primaryKey(sortKeyFields: ["warehouseID"])
  warehouseID: ID!
  InventoryAmount: Int!
}

The schema above will allow you to pass different conditions to query the correct inventory item:

query QueryInventoryByProductAndWarehouse($productID: ID!, $warehouseID: ID!) {
  getInventory(productID: $productID, warehouseID: $warehouseID) {
    productID
    warehouseID
    inventoryAmount
  }
}

import { getInventory } from './graphql/queries';

const result = await client.graphql({
  query: getInventory,
  variables: {
    productID: 'product-id',
    warehouseID: 'warehouse-id'
  }
});
const inventory = result.data.getInventory;

Configure a secondary index

Amplify uses Amazon DynamoDB tables as the underlying data source for @model types. For key-value databases, it is critical to model your access patterns with "secondary indexes". Use the @index directive to configure a secondary index.

Amazon DynamoDB is a key-value and document database that delivers single-digit millisecond performance at any scale but making it work for your access patterns requires a bit of forethought. DynamoDB query operations may use at most two attributes to efficiently query data. The first query argument passed to a query (the hash key) must use strict equality and the second attribute (the sort key) may use gt, ge, lt, le, eq, beginsWith, and between. DynamoDB can effectively implement a wide variety of access patterns that are powerful enough for the majority of applications.

A secondary index consists of a "hash key" and, optionally, a "sort key". Use the "hash key" to perform strict equality and the "sort key" for greater than (gt), greater than or equal to (ge), less than (lt), less than or equal to (le), equals (eq), begins with, and between operations.

type Customer @model {
  id: ID!
  name: String!
  phoneNumber: String
  accountRepresentativeID: ID! @index
}

The example client query below allows you to query for "Customer" records based on their accountRepresentativeID:

query QueryCustomersForAccountRepresentative($accountRepresentativeID: ID!) {
  customersByAccountRepresentativeID(
    accountRepresentativeID: $accountRepresentativeID
  ) {
    customers {
      items {
        id
        name
        phoneNumber
      }
    }
  }
}

import { customersByAccountRepresentativeID } from './graphql/queries';

const result = await client.graphql({
  query: customersByAccountRepresentativeID,
  variables: {
    accountRepresentativeID: 'account-rep-id'
  }
});
const customers = result.data.customersByAccountRepresentativeID;

You can also overwrite the queryField or name to customize the GraphQL query name, or secondary index name respectively:

type Customer @model {
  id: ID!
  name: String!
  phoneNumber: String
  accountRepresentativeID: ID!
    @index(name: "byRepresentative", queryField: "customerByRepresentative")
}

query QueryCustomersForAccountRepresentative($representativeId: ID!) {
  customerByRepresentative(accountRepresentativeID: $representativeId) {
    customers {
      items {
        id
        name
        phoneNumber
      }
    }
  }
}

import { customerByRepresentative } from './graphql/queries';

const result = await client.graphql({
  query: customerByRepresentative,
  variables: {
    accountRepresentativeID: 'account-rep-id'
  }
});
const customer = result.data.customerByRepresentative;

To optionally configure sort keys, provide the additional fields in the sortKeyFields parameter:

type Customer @model @auth(rules: [{ allow: public }]) {
  id: ID!
  name: String! @index(name: "byNameAndPhoneNumber", sortKeyFields: ["phoneNumber"], queryField: "customerByNameAndPhone")
  phoneNumber: String
  accountRepresentativeID: ID! @index

The example client query below allows you to query for "Customer" based on their name and filter based on phoneNumber:

query MyQuery {
  customerByNameAndPhone(phoneNumber: { beginsWith: "+1" }, name: "Rene") {
    items {
      id
      name
      phoneNumber
    }
  }
}

import { customerByNameAndPhone } from './graphql/queries';

const result = await client.graphql({
  query: customerByNameAndPhone,
  variables: {
    phoneNumber: { beginsWith: '+1' },
    name: 'Rene'
  }
});

const customer = result.data.customerByNameAndPhone;

Setup relationships between models

Create "has one", "has many", "belongs to", and "many to many" relationships between @model types.

Has One relationship

Create a one-directional one-to-one relationship between two models using the @hasOne directive.

In the example below, a Project has a Team.

type Project @model {
  id: ID!
  name: String
  team: Team @hasOne
}

type Team @model {
  id: ID!
  name: String!
}

This generates queries and mutations that allow you to retrieve the related record from the source record:

mutation CreateProject {
  createProject(input: { projectTeamId: "team-id", name: "Some Name" }) {
    team {
      name
      id
    }
    name
    id
  }
}

import { createProject } from './graphql/mutations';

const result = await client.graphql({
  query: createProject,
  variables: {
    input: { projectTeamId: 'team-id', name: 'Some Name' }
  }
});

const project = result.data.createProject;

To customize the field to be used for storing the relationship information, set the fields array argument and matching it to a field on the type:

type Project @model {
  id: ID!
  name: String
  teamID: ID
  team: Team @hasOne(fields: ["teamID"])
}

type Team @model {
  id: ID!
  name: String!
}

In this case, the Project type has a teamID field added as an identifier for the team. @hasOne can then get the connected Team object by querying the Team table with this teamID:

mutation CreateProject {
  createProject(input: { name: "New Project", teamID: "a-team-id" }) {
    id
    name
    team {
      id
      name
    }
  }
}

import { createProject } from './graphql/mutations';

const result = await client.graphql({
  query: createProject,
  variables: {
    input: {
      teamID: 'team-id',
      name: 'New Project'
    }
  }
});
const project = result.data.createProject;

A @hasOne relationship always uses a reference to the primary key of the related model, by default id unless overridden with the @primaryKey directive.

Has Many relationship

Create a one-directional one-to-many relationship between two models using the @hasMany directive.

type Post @model {
  id: ID!
  title: String!
  comments: [Comment] @hasMany
}

type Comment @model {
  id: ID!
  content: String!
}

This generates queries and mutations that allow you to retrieve the related Comment records from the source Post record:

mutation CreatePost {
  createPost(input: { title: "Hello World!!" }) {
    title
    id
    comments {
      items {
        id
        content
      }
    }
  }
}

import { createPost } from './graphql/mutations';

const result = await client.graphql({
  query: createPost,
  variables {
    input: { title: 'Hello World!!' },
  }
});
const post = result.data.createPost;
const comments = post.comments.items;

Under the hood, @hasMany configures a default secondary index on the related table to enable you to query the related model from the source model.

To customize the specific secondary index used for the "has many" relationship, create an @index directive on the field in the related table where you want to assign the secondary index.

Next, provide the secondary index with a name attribute and a value. Optionally, you can configure a “sort key” on the secondary index by providing a sortKeyFields attribute and a designated field as its value.

On the @hasMany configuration, pass in the name value from your secondary index as the value for the indexName parameter. Then, pass in the respective fields that match the connected index.

type Post @model {
  id: ID!
  title: String!
  comments: [Comment] @hasMany(indexName: "byPost", fields: ["id"])
}

type Comment @model {
  id: ID!
  postID: ID! @index(name: "byPost", sortKeyFields: ["content"])
  content: String!
}

In this case, the Comment type has a postID field added to store the reference of Post record. The id field referenced by @hasMany is the id on the Post type. @hasMany can then get the connected Comment object by querying the Comment table's secondary index "byPost" with this postID:

mutation CreatePost {
  createPost(input: { title: "Hello world!" }) {
    comments {
      items {
        postID
        content
        id
      }
    }
    title
    id
  }
}

import { createPost, createComment } from './graphql/mutations';
import { getPost } from './graphql/mutations';

// create post
const result = await client.graphql({
  query: createPost,
  variables: {
    input: { title: 'Hello World!!' }
  }
});
const post = result.data.createPost;

// create comment
await client.graphql({
  query: createComment,
  variables: {
    input: { content: 'Hi!', postID: post.id }
  }
});

// get post
const result = await client.graphql({
  query: getPost,
  variables: { id: post.id }
});

const postWithComments = result.data.createPost;
const postComments = postWithComments.comments.items; // access comments from post

Belongs To relationship

Make a "has one" or "has many" relationship bi-directional with the @belongsTo directive.

type Project @model {
  id: ID!
  name: String
  team: Team @hasOne
}

type Team @model {
  id: ID!
  name: String!
  project: Project @belongsTo
}

mutation CreateProject {
  createProject(input: { name: "New Project", teamID: "a-team-id" }) {
    id
    name
    team {
      # query team from project
      id
      name
      project {
        # bi-directional query: team to project
        id
        name
      }
    }
  }
}

import { createProject, createTeam, updateTeam } from './graphql/mutations';

// create team
const result = await client.graphql({
  query: createTeam,
  variables: {
    input: { name: 'New Team' }
  }
});
const team = result.data.createTeam;

// create project
const result = await client.graphql({
  query: createProject,
  variables: {
    input: { name: 'New Project', projectTeamId: team.id }
  }
});
const project = result.data.createProject;
const projectTeam = project.team; // access team from project

// update team
const updateTeamResult = await client.graphql({
  query: updateTeam,
  variables: {
    input: { id: team.id, teamProjectId: project.id }
  }
});

const updatedTeam = updateTeamResult.data.updateTeam;
const teamProject = updatedTeam.project; // access project from team

type Post @model {
  id: ID!
  title: String!
  comments: [Comment] @hasMany
}

type Comment @model {
  id: ID!
  content: String!
  post: Post @belongsTo
}

mutation CreatePost {
  createPost(input: { title: "Hello World!!" }) {
    title
    id
    comments {
      # query comments from the post
      items {
        id
        content
        post {
          # bi-directional query: comment to post
          id
          title
        }
      }
    }
  }
}

import { createPost, createComment } from './graphql/mutations';
import { getPost } from './graphql/mutations';

// create post
const result = await client.graphql({
  query: createPost,
  variables: {
    input: { title: 'Hello World!!' }
  }
});
const post = result.data.createPost;

// create comment
await client.graphql({
  query: createComment,
  variables: {
    input: { content: 'Hi!', postID: post.id }
  }
});

// get post
const result = await client.graphql({
  query: getPost,
  variables: { id: post.id }
});

const postWithComments = result.data.createPost;
const postComments = postWithComments.comments.items; // access comments from post

const commentPost = postComments[0].post; // access post from comment;

@belongsTo can be used without the fields argument. In those cases, a field is automatically generated to reference the parent’s primary key.

Alternatively, you set up a custom field to store the reference of the parent object. An example bidirectional “has many” relationship is shown below.

type Post @model {
  id: ID!
  title: String!
  comments: [Comment] @hasMany(indexName: "byPost", fields: ["id"])
}

type Comment @model {
  id: ID!
  postID: ID! @index(name: "byPost", sortKeyFields: ["content"])
  content: String!
  post: Post @belongsTo(fields: ["postID"])
}

Note: The @belongsTo directive requires that a @hasOne or @hasMany relationship already exists from parent to the related model.

Many-to-many relationship

Create a many-to-many relationship between two models with the @manyToMany directive. Provide a common relationName on both models to join them into a many-to-many relationship.

type Post @model {
  id: ID!
  title: String!
  content: String
  tags: [Tag] @manyToMany(relationName: "PostTags")
}

type Tag @model {
  id: ID!
  label: String!
  posts: [Post] @manyToMany(relationName: "PostTags")
}

Under the hood, the @manyToMany directive will create a "join table" named after the relationName to facilitate the many-to-many relationship. This generates queries and mutations that allow you to retrieve the related Comment records from the source Post record and vice versa:

mutation CreatePost {
  createPost(input: { title: "Hello World!!" }) {
    id
    title
    content
    tags {
      # queries the "join table" PostTags
      items {
        tag {
          # related Tag records from Post
          id
          label
          posts {
            # queries the "join table" PostTags
            items {
              post {
                # related Post records from Tag
                id
                title
                content
              }
            }
          }
        }
      }
    }
  }
}

import { createPost, createTag, createPostTags } from './graphql/mutations';
import { listPosts } from './graphql/queries';

// create post
const result = await client.graphql({
  query: createPost,
  variables: {
    input: { title: 'Hello World' }
  }
});
const post = result.data.createPost;

// create tag
const tagResult = await client.graphql({
  query: createTag,
  variables: {
    input: {
      label: 'My Tag'
    }
  }
});
const tag = tagResult.data.createTag;

// connect post and tag
await client.graphql({
  query: createPostTags,
  variables: {
    input: {
      postId: post.id,
      tagId: tag.id
    }
  }
});

// get posts
const listPostsResult = await client.graphql({ query: listPosts });
const posts = listPostsResult.data.listPosts;

const postTags = posts[0].tags; // access tags from post

Assign default values for fields

You can use the @default directive to specify a default value for optional scalar type fields such as Int, String, and more.

type Todo @model {
  content: String @default(value: "My new Todo")
  # Note: all "value" parameters must be passed as a string value. 
  # Under the hood, Amplify will parse the string values into respective types.
  # For example, to set a default value for an integer field, 
  # you must pass in `"0"` instead of `0` without the double-quotes.
  likes: Int @default(value: "0") # 
}

If you create a new Todo and don't supply a content input, Amplify will ensure that My new Todo is auto populated as a value. When @default is applied, non-null assertions using ! are disregarded. For example, String! is treated the same as String.

Server-side filtering for subscriptions

A server-side subscription filter expression is automatically generated for any @model type.

type Task @model {
  title: String!
  description: String
  type: String
  priority: Int
}

You can filter the subscriptions server-side by passing a filter expression. For example: If you want to subscribe to tasks of type Security and priority greater than 5, you can set the filter argument accordingly.

subscription OnCreateTask {
  onCreateTask(
    filter: { and: [{ type: { eq: "Security" } }, { priority: { gt: 5 } }] }
  ) {
    title
    description
    type
    priority
  }
}

import { onCreateTask } from './graphql/subscriptions';

const subscription = client.graphql({
  query: onCreateTask,
  variables: {
    filter: {
      and: [
        { type: { eq: "Security" } }
        { priority: { gt: 5 } }
      ]
    }
  }
}).subscribe({
	next: ({ data }) => console.log(data),
	error: (error) => console.warn(error)
});

If you want to get all subscription events, don’t pass any filter parameters.

Advanced

Rename generated queries, mutations, and subscriptions

You can override the names of any @model-generated GraphQL queries, mutations, and subscriptions by supplying the desired name.

type Todo @model(queries: { get: "queryFor" }) {
  name: String!
  description: String
}

In the example above, you will be able to run a queryForTodo query to get a single Todo element.

Disable generated queries, mutations, and subscriptions

You can disable specific operations by assigning their value to null.

type Todo @model(queries: { get: null }, mutations: null, subscriptions: null) {
  name: String!
  description: String
}

The example above disables the getTodo query, all mutations, and all subscriptions while allowing the generation of other queries such as listTodo.

Creating a custom query

You can disable the get query and create a custom query that enables us to retrieve a single Todo model.

type Query {
  getMyTodo(id: ID!): Todo @function(name: "getmytodofunction-${env}")
}

The example above creates a custom query that utilizes the @function directive to call a Lambda function for this query.

For the type definitions of queries, mutations, and subscriptions, see Type Definitions of the @model Directive.

Customize creation and update timestamps

The @model directive automatically adds createdAt and updatedAt timestamps to each entity. The timestamp field names can be changed by passing timestamps attribute to the directive.

type Todo
  @model(timestamps: { createdAt: "createdOn", updatedAt: "updatedOn" }) {
  name: String!
  description: String
}

For example, the schema above will allow you to query for the following contents:

type Todo {
  id: ID!
  name: String!
  description: String
  createdOn: AWSDateTime!
  updatedOn: AWSDateTime!
}

Modify subscriptions (real-time updates) access level

By default, real-time updates are on for all @model types, which means customers receive real-time updates and authorization rules are applied during initial connection time. You can also turn off subscriptions for that model or make the real-time updates public, receivable by all subscribers.

type Todo
  @model(subscriptions: { level: off }) { # or level: public
  name: String!
  description: String
}

Create multiple relationships between two models

You need to explicitly specify the connection field names if relational directives are used to create two connections of the same type between the two models.

type Individual @model {
  id: ID!
  homeAddress: Address @hasOne
  shippingAddress: Address @hasOne
}

type Address @model {
  id: ID!
  homeIndividualID: ID
  shippingIndividualID: ID
  homeIndividual: Individual @belongsTo(fields: ["homeIndividualID"])
  shipIndividual: Individual @belongsTo(fields: ["shippingIndividualID"])
}

Relationships to a model with a composite primary key

When a primary key is defined by a sort key in addition to the hash key, then it's called a composite primary key.

If you explicitly define the fields argument on the @hasOne, @hasMany, or @belongsTo directives and reference a model that has a composite primary key, then you must set the values in the fields argument in a specific order:

• The first value should always be the primary key of the related model.
• Remaining values should match the sortKeyFields specified in the @primaryKey directive of the related model.

type Project @model {
  projectId: ID! @primaryKey(sortKeyFields: ["name"])
  name: String!
  team: Team @hasOne(fields: ["teamId", "teamName"])
  teamId: ID # customized foreign key for child primary key
  teamName: String # customized foreign key for child sort key
}

type Team @model {
  teamId: ID! @primaryKey(sortKeyFields: ["name"])
  name: String!
}

type Project @model {
  projectId: ID! @primaryKey(sortKeyFields: ["name"])
  name: String!
  team: Team @hasOne(fields: ["teamId", "teamName"])
  teamId: ID # customized foreign key for child primary key
  teamName: String # customized foreign key for child sort key
}

type Team @model {
  teamId: ID! @primaryKey(sortKeyFields: ["name"])
  name: String!
  project: Project @belongsTo(fields: ["projectId", "projectName"])
  projectId: ID # customized foreign key for parent primary key
  projectName: String # customized foreign key for parent sort key
}

type Post @model {
  postId: ID! @primaryKey(sortKeyFields: ["title"])
  title: String!
  comments: [Comment] @hasMany(indexName: "byPost", fields: ["postId", "title"])
}

type Comment @model {
  commentId: ID! @primaryKey(sortKeyFields: ["content"])
  content: String!
  postId: ID @index(name: "byPost", sortKeyFields: ["postTitle"]) # customized foreign key for parent primary key
  postTitle: String # customized foreign key for parent sort key
}

type Post @model {
  postId: ID! @primaryKey(sortKeyFields: ["title"])
  title: String!
  comments: [Comment] @hasMany(indexName: "byPost", fields: ["postId", "title"])
}

type Comment @model {
  commentId: ID! @primaryKey(sortKeyFields: ["content"])
  content: String!
  post: Post @belongsTo(fields: ["postId", "postTitle"])
  postId: ID @index(name: "byPost", sortKeyFields: ["postTitle"]) # customized foreign key for parent primary key
  postTitle: String # customized foreign key for parent sort key
}

Generate a secondary index without a GraphQL query

Because query creation against a secondary index is automatic, if you wish to define a secondary index that does not have a corresponding query in your API, set the queryField parameter to null.

type Customer @model {
  id: ID!
  name: String!
  phoneNumber: String
  accountRepresentativeID: ID! @index(queryField: null)
}

Split GraphQL files

AWS Amplify supports splitting your GraphQL schema into separate .graphql files.

You can start by creating a amplify/backend/api/<api-name>/schema/ directory. As an example, you might split up the schema for a blog site by creating Blog.graphql, Post.graphql, and Comment.graphql files.

You can then run amplify api gql-compile and the output build schema will include all the types declared across your schema files.

As your project grows, you may want to organize your custom queries, mutations, and subscriptions depending on the size and maintenance requirements of your project. You can either consolidate all of them into one file or colocate them with their corresponding models.

Using a Single Query.graphql File

This method involves consolidating all queries into a single Query.graphql file. It is useful for smaller projects or when you want to keep all queries in one place.

1. In the amplify/backend/api/<api-name>/schema/ directory, create a file named Query.graphql.

2. Copy all query type definitions from your multiple schema files into the Query.graphql file.

3. Make sure all your queries are properly formatted and enclosed within a single type Query { ... } block.

Using the extend Keyword

Declaring a Query type in separate schema files will result in schema validation errors similar to the following when running amplify api gql-compile:

🛑 Schema validation failed.

There can be only one type named "Query".

Amplify GraphQL schemas support the extend keyword, which allows you to extend types with additional fields. In this case, it also allows you to split your custom queries, mutations, and subscriptions into multiple files. This may be more ideal for larger, more complex projects.

1. Organize your GraphQL schema into multiple files as per your project's architecture.

2. In one of the files (e.g., schema1.graphql), declare your type normally:

type Query {
  # initial custom queries
}

3. In other schema files (e.g., schema2.graphql), use the extend keyword to add to the type:

extend type Query {
  # additional custom queries
}

How it works

Model directive

The @model directive will generate:

• An Amazon DynamoDB table with PAY_PER_REQUEST billing mode enabled by default.
• An AWS AppSync DataSource configured to access the table above.
• An AWS IAM role attached to the DataSource that allows AWS AppSync to call the above table on your behalf.
• Up to 8 resolvers (create, update, delete, get, list, onCreate, onUpdate, onDelete) but this is configurable via the queries, mutations, and subscriptions arguments on the @model directive.
• Input objects for create, update, and delete mutations.
• Filter input objects that allow you to filter objects in list queries and relationship fields.
• For list queries the default number of objects returned is 100. You can override this behavior by setting the limit argument.

Type definition of the @model directive

directive @model(
  queries: ModelQueryMap
  mutations: ModelMutationMap
  subscriptions: ModelSubscriptionMap
  timestamps: TimestampConfiguration
) on OBJECT

input ModelMutationMap {
  create: String
  update: String
  delete: String
}

input ModelQueryMap {
  get: String
  list: String
}

input ModelSubscriptionMap {
  onCreate: [String]
  onUpdate: [String]
  onDelete: [String]
  level: ModelSubscriptionLevel
}

enum ModelSubscriptionLevel {
  off
  public
  on
}

input TimestampConfiguration {
  createdAt: String
  updatedAt: String
}

Relational directives

The relational directives are @hasOne, @hasMany, @belongsTo and @manyToMany.

directive @hasOne(fields: [String!]) on FIELD_DEFINITION

directive @hasMany(
  indexName: String
  fields: [String!]
  limit: Int = 100
) on FIELD_DEFINITION

directive @belongsTo(fields: [String!]) on FIELD_DEFINITION

directive @manyToMany(
  relationName: String!
  limit: Int = 100
) on FIELD_DEFINITION