mongodb is the official low-level driver for MongoDB, offering direct access to database commands without abstraction. mongoose is an Object Data Modeling (ODM) library built on top of mongodb, providing schema validation and middleware. sequelize is a promise-based ORM for SQL databases like PostgreSQL and MySQL, focusing on ease of use. typeorm is an ORM that supports both SQL and MongoDB, emphasizing TypeScript decorators and entity-based architecture.
Selecting the right data layer is a critical architectural decision for Node.js applications. mongodb, mongoose, sequelize, and typeorm each solve data persistence differently β ranging from raw drivers to full-featured ORMs. Understanding their operational models helps you avoid bottlenecks and maintenance issues down the line.
mongodb requires you to manage the client connection manually.
MongoClient and connect to the URI.// mongodb: Manual client connection
const { MongoClient } = require('mongodb');
const client = new MongoClient('mongodb://localhost:27017');
await client.connect();
const db = client.db('myApp');
mongoose wraps the connection logic in a singleton pattern.
mongodb client for you.mongoose.connection.// mongoose: Managed connection
const mongoose = require('mongoose');
await mongoose.connect('mongodb://localhost:27017/myApp');
// Access via mongoose.connection
sequelize initializes a connection instance for SQL databases.
// sequelize: SQL connection instance
const { Sequelize } = require('sequelize');
const sequelize = new Sequelize('postgres://user:pass@localhost:5432/db');
await sequelize.authenticate();
typeorm uses a DataSource object to manage connections.
// typeorm: DataSource configuration
const { DataSource } = require('typeorm');
const AppDataSource = new DataSource({
type: 'postgres',
host: 'localhost',
port: 5432,
entities: [User]
});
await AppDataSource.initialize();
mongodb has no schema enforcement at the driver level.
// mongodb: Raw document insertion
const users = db.collection('users');
await users.insertOne({ name: 'Alice', age: 30 });
// No schema validation here
mongoose enforces structure using Schemas.
// mongoose: Schema definition
const userSchema = new mongoose.Schema({
name: String,
age: Number
});
const User = mongoose.model('User', userSchema);
sequelize defines models that map to SQL tables.
// sequelize: Model definition
const User = sequelize.define('User', {
name: { type: Sequelize.STRING },
age: { type: Sequelize.INTEGER }
});
await sequelize.sync();
typeorm uses classes and decorators to define entities.
// typeorm: Entity class
@Entity()
class User {
@PrimaryGeneratedColumn()
id: number;
@Column()
name: string;
}
mongodb uses standard database commands for queries.
// mongodb: Raw query
const user = await db.collection('users').findOne({ age: 30 });
mongoose provides a chainable query API.
// mongoose: Chainable query
const user = await User.findOne({ age: 30 });
sequelize offers a robust query builder.
findAll or findOne.// sequelize: Query builder
const user = await User.findOne({ where: { age: 30 } });
typeorm relies on repositories or query builders.
// typeorm: Repository find
const repo = AppDataSource.getRepository(User);
const user = await repo.findOne({ where: { age: 30 } });
mongodb handles relationships via manual references or embedding.
$lookup).// mongodb: Manual reference
const post = await db.collection('posts').findOne({ authorId: user._id });
mongoose supports population to resolve references.
ref in the schema.// mongoose: Population
const post = await Post.findById(id).populate('author');
sequelize excels at SQL joins and associations.
hasMany or belongsTo relationships.// sequelize: Associations
const user = await User.findByPk(id, { include: Post });
typeorm manages relations via entity decorators.
@OneToMany or @ManyToOne.// typeorm: Relations
@OneToMany(() => Post, post => post.author)
posts: Post[];
While the underlying databases differ, these libraries share common goals and patterns.
await directly without callback hell.// All packages support this pattern
async function getData() {
const result = await libraryQuery();
return result;
}
// Standard error handling
try {
await dbOperation();
} catch (error) {
console.error(error);
}
// Implicit in all libraries
// No need to open/close connection per query
| Feature | mongodb | mongoose | sequelize | typeorm |
|---|---|---|---|---|
| Database | MongoDB | MongoDB | SQL (Postgres, MySQL) | SQL & MongoDB |
| Abstraction | None (Driver) | ODM | ORM | ORM |
| Schema | None | Enforced | Enforced | Enforced (Classes) |
| Query Style | Raw Commands | Chainable API | Query Builder | Repository/Builder |
| TypeScript | Manual Types | Built-in Types | Good Support | Excellent Support |
| Feature | Shared by All |
|---|---|
| Async Model | β Promises / Async-Await |
| Connection Mgmt | β Pooling Included |
| Error Handling | β Try/Catch Compatible |
| Ecosystem | β Large Community Support |
mongodb is like driving a manual transmission car π β you have full control over every gear shift, but you must manage the mechanics yourself. Best for experts who need raw speed and flexibility.
mongoose is like an automatic car with safety features π‘οΈ β it handles the shifting (schema validation) for you while keeping you on the MongoDB road. Ideal for teams wanting structure without leaving NoSQL.
sequelize is like a reliable truck π β built for heavy lifting with SQL databases, offering stable associations and migrations. Perfect for traditional relational data needs.
typeorm is like a modern electric vehicle with smart sensors π€ β it uses TypeScript decorators to automate much of the wiring. Best for TypeScript-heavy projects needing type safety.
Final Thought: Your choice depends on your data shape and team skills. If you need flexible documents, pick mongoose. If you need strict relations, pick sequelize or typeorm. If you need raw performance without abstraction, pick mongodb.
Choose mongodb if you need maximum control over database operations and want to avoid abstraction layers. It is ideal for high-performance scenarios where you need to execute specific MongoDB commands directly without overhead. This package suits teams comfortable managing schema validation manually in application code.
Choose mongoose if you are using MongoDB and want built-in schema validation, casting, and middleware. It simplifies data modeling for teams who prefer structure over raw flexibility. This package is best for applications that benefit from defined data shapes without switching to a SQL database.
Choose sequelize if you are working with SQL databases and prefer a promise-based API with strong support for migrations. It is well-suited for teams that want a mature, stable ORM with extensive documentation for relational data. This package works well for traditional CRUD applications requiring complex joins.
Choose typeorm if you are using TypeScript and want an ORM that leverages decorators for entity definition. It is ideal for projects that value type safety and might need to switch between SQL and MongoDB drivers later. This package fits architectures that rely heavily on class-based design patterns.
The official MongoDB driver for Node.js.
Upgrading to version 7? Take a look at our upgrade guide here!
| Site | Link |
|---|---|
| Documentation | www.mongodb.com/docs/drivers/node |
| API Docs | mongodb.github.io/node-mongodb-native |
npm package | www.npmjs.com/package/mongodb |
| MongoDB | www.mongodb.com |
| MongoDB University | learn.mongodb.com |
| MongoDB Developer Center | www.mongodb.com/developer |
| Stack Overflow | stackoverflow.com |
| Source Code | github.com/mongodb/node-mongodb-native |
| Upgrade to v7 | etc/notes/CHANGES_7.0.0.md |
| Contributing | CONTRIBUTING.md |
| Changelog | HISTORY.md |
Releases are created automatically and signed using the Node team's GPG key. This applies to the git tag as well as all release packages provided as part of a GitHub release. To verify the provided packages, download the key and import it using gpg:
gpg --import node-driver.asc
The GitHub release contains a detached signature file for the NPM package (named
mongodb-X.Y.Z.tgz.sig).
The following command returns the link npm package.
npm view mongodb@vX.Y.Z dist.tarball
Using the result of the above command, a curl command can return the official npm package for the release.
To verify the integrity of the downloaded package, run the following command:
gpg --verify mongodb-X.Y.Z.tgz.sig mongodb-X.Y.Z.tgz
[!Note] No verification is done when using npm to install the package. The contents of the Github tarball and npm's tarball are identical.
The MongoDB Node.js driver follows semantic versioning for its releases.
Think youβve found a bug? Want to see a new feature in node-mongodb-native? Please open a
case in our issue management tool, JIRA:
Bug reports in JIRA for all driver projects (i.e. NODE, PYTHON, CSHARP, JAVA) and the Core Server (i.e. SERVER) project are public.
For issues with, questions about, or feedback for the Node.js driver, please look into our support channels. Please do not email any of the driver developers directly with issues or questions - you're more likely to get an answer on the MongoDB Community Forums.
Change history can be found in HISTORY.md.
The driver currently supports 4.2+ servers.
For exhaustive server and runtime version compatibility matrices, please refer to the following links:
The following table describes add-on component version compatibility for the Node.js driver. Only packages with versions in these supported ranges are stable when used in combination.
| Component | mongodb@3.x | mongodb@4.x | mongodb@5.x | mongodb@<6.12 | mongodb@>=6.12 | mongodb@7.x |
|---|---|---|---|---|---|---|
| bson | ^1.0.0 | ^4.0.0 | ^5.0.0 | ^6.0.0 | ^6.0.0 | ^7.0.0 |
| bson-ext | ^1.0.0 || ^2.0.0 | ^4.0.0 | N/A | N/A | N/A | N/A |
| kerberos | ^1.0.0 | ^1.0.0 || ^2.0.0 | ^1.0.0 || ^2.0.0 | ^2.0.1 | ^2.0.1 | ^7.0.0 |
| mongodb-client-encryption | ^1.0.0 | ^1.0.0 || ^2.0.0 | ^2.3.0 | ^6.0.0 | ^6.0.0 | ^7.0.0 |
| mongodb-legacy | N/A | ^4.0.0 | ^5.0.0 | ^6.0.0 | ^6.0.0 | N/A |
| @mongodb-js/zstd | N/A | ^1.0.0 | ^1.0.0 | ^1.1.0 | ^1.1.0 || ^2.0.0 | ^7.0.0 |
We recommend using the latest version of typescript, however we currently ensure the driver's public types compile against typescript@5.6.0.
This is the lowest typescript version guaranteed to work with our driver: older versions may or may not work - use at your own risk.
Since typescript does not restrict breaking changes to major versions, we consider this support best effort.
If you run into any unexpected compiler failures against our supported TypeScript versions, please let us know by filing an issue on our JIRA.
Additionally, our Typescript types are compatible with the ECMAScript standard for our minimum supported Node version. Currently, our Typescript targets es2023.
The recommended way to get started using the Node.js driver is by using the npm (Node Package Manager) to install the dependency in your project.
After you've created your own project using npm init, you can run:
npm install mongodb
This will download the MongoDB driver and add a dependency entry in your package.json file.
If you are a Typescript user, you will need the Node.js type definitions to use the driver's definitions:
npm install -D @types/node
The MongoDB driver can optionally be enhanced by the following feature packages:
Maintained by MongoDB:
Some of these packages include native C++ extensions. Consult the trouble shooting guide here if you run into compilation issues.
Third party:
This guide will show you how to set up a simple application using Node.js and MongoDB. Its scope is only how to set up the driver and perform the simple CRUD operations. For more in-depth coverage, see the official documentation.
package.json fileFirst, create a directory where your application will live.
mkdir myProject
cd myProject
Enter the following command and answer the questions to create the initial structure for your new project:
npm init -y
Next, install the driver as a dependency.
npm install mongodb
For complete MongoDB installation instructions, see the manual.
mongod process.mongod --dbpath=/data
You should see the mongod process start up and print some status information.
Create a new app.js file and add the following code to try out some basic CRUD operations using the MongoDB driver.
Add code to connect to the server and the database myProject:
NOTE: Resolving DNS Connection issues
Node.js 18 changed the default DNS resolution ordering from always prioritizing IPv4 to the ordering returned by the DNS provider. In some environments, this can result in
localhostresolving to an IPv6 address instead of IPv4 and a consequent failure to connect to the server.This can be resolved by:
- specifying the IP address family using the MongoClient
familyoption (MongoClient(<uri>, { family: 4 } ))- launching mongod or mongos with the ipv6 flag enabled (--ipv6 mongod option documentation)
- using a host of
127.0.0.1in place of localhost- specifying the DNS resolution ordering with the
--dns-resolution-orderNode.js command line argument (e.g.node --dns-resolution-order=ipv4first)
const { MongoClient } = require('mongodb');
// or as an es module:
// import { MongoClient } from 'mongodb'
// Connection URL
const url = 'mongodb://localhost:27017';
const client = new MongoClient(url);
// Database Name
const dbName = 'myProject';
async function main() {
// Use connect method to connect to the server
await client.connect();
console.log('Connected successfully to server');
const db = client.db(dbName);
const collection = db.collection('documents');
// the following code examples can be pasted here...
return 'done.';
}
main()
.then(console.log)
.catch(console.error)
.finally(() => client.close());
Run your app from the command line with:
node app.js
The application should print Connected successfully to server to the console.
Add to app.js the following function which uses the insertMany method to add three documents to the documents collection.
const insertResult = await collection.insertMany([{ a: 1 }, { a: 2 }, { a: 3 }]);
console.log('Inserted documents =>', insertResult);
The insertMany command returns an object with information about the insert operations.
Add a query that returns all the documents.
const findResult = await collection.find({}).toArray();
console.log('Found documents =>', findResult);
This query returns all the documents in the documents collection. If you add this below the insertMany example, you'll see the documents you've inserted.
Add a query filter to find only documents which meet the query criteria.
const filteredDocs = await collection.find({ a: 3 }).toArray();
console.log('Found documents filtered by { a: 3 } =>', filteredDocs);
Only the documents which match 'a' : 3 should be returned.
The following operation updates a document in the documents collection.
const updateResult = await collection.updateOne({ a: 3 }, { $set: { b: 1 } });
console.log('Updated documents =>', updateResult);
The method updates the first document where the field a is equal to 3 by adding a new field b to the document set to 1. updateResult contains information about whether there was a matching document to update or not.
Remove the document where the field a is equal to 3.
const deleteResult = await collection.deleteMany({ a: 3 });
console.log('Deleted documents =>', deleteResult);
Indexes can improve your application's performance. The following function creates an index on the a field in the documents collection.
const indexName = await collection.createIndex({ a: 1 });
console.log('index name =', indexName);
For more detailed information, see the indexing strategies page.
If you need to filter certain errors from our driver, we have a helpful tree of errors described in etc/notes/errors.md.
It is our recommendation to use instanceof checks on errors and to avoid relying on parsing error.message and error.name strings in your code.
We guarantee instanceof checks will pass according to semver guidelines, but errors may be sub-classed or their messages may change at any time, even patch releases, as we see fit to increase the helpfulness of the errors.
Any new errors we add to the driver will directly extend an existing error class and no existing error will be moved to a different parent class outside of a major release.
This means instanceof will always be able to accurately capture the errors that our driver throws.
const client = new MongoClient(url);
await client.connect();
const collection = client.db().collection('collection');
try {
await collection.insertOne({ _id: 1 });
await collection.insertOne({ _id: 1 }); // duplicate key error
} catch (error) {
if (error instanceof MongoServerError) {
console.log(`Error worth logging: ${error}`); // special case for some reason
}
throw error; // still want to crash
}
If you need to test with a change from the latest main branch, our mongodb npm package has nightly versions released under the nightly tag.
npm install mongodb@nightly
Nightly versions are published regardless of testing outcome. This means there could be semantic breakages or partially implemented features. The nightly build is not suitable for production use.
Β© 2012-present MongoDB Contributors
Β© 2009-2012 Christian Amor Kvalheim