Backend6 min read

MongoDB

Document-oriented NoSQL database - flexible schemas, horizontal scaling, and rich aggregation pipeline make it ideal for hierarchical or rapidly evolving data models.

MongoDBNoSQLMongooseAggregationAtlas

skill: MongoDB category: Backend tags: [MongoDB, NoSQL, Mongoose, Aggregation, Atlas] readingTime: 6

MongoDB is ideal for hierarchical data (nested documents), rapidly changing schemas, and horizontal scaling. For relational data with strong consistency requirements, use PostgreSQL. For time-series data, use InfluxDB or TimescaleDB. Don't default to MongoDB just because it's 'easier' - schema discipline matters.

Workflow

Connect and model with MongooseMongoose adds schemas, validation, and middleware on top of the MongoDB driver - essential for enforcing data consistency in a schemaless DB.
CRUD operationsMongoDB's query language is JSON-based - intuitive for JavaScript developers and powerful for nested document queries.
Use the aggregation pipelineThe aggregation pipeline is MongoDB's SQL GROUP BY + JOIN - transform, group, and analyze documents in one query.
Index strategicallyMongoDB queries without an index do a full collection scan - fine for small collections, catastrophic at scale.
Use Atlas Search for full-text searchAtlas Search (Lucene-based) handles fuzzy matching, typo tolerance, synonyms, and relevance ranking - far beyond MongoDB's native text indexes.

Output Format

MongoDB data layer with Mongoose schemas, indexed queries, and aggregation pipelines.

models/
├── User.ts        # Schema + indexes + methods
├── Post.ts
└── Order.ts

repositories/
├── userRepo.ts    # CRUD + aggregation queries
└── orderRepo.ts

Examples

Input

Output

Edge Cases & Gotchas

ObjectId vs. string IDs - MongoDB uses ObjectId for _id by default. When querying by _id, pass new mongoose.Types.ObjectId(id) - passing a raw string won't match.
Mongoose validation on update - Model.updateOne() bypasses Mongoose validators by default. Use { runValidators: true } in options.
Connection pool exhaustion - Too many concurrent operations exhaust the connection pool. Set maxPoolSize=50 (default 5) and add connection monitoring.
Array mutation operators - Replacing an array with $set replaces the entire array. Use $push, $pull, $addToSet for mutations that preserve other elements.

When to Break These Rules

Use PostgreSQL when data is highly relational, requires complex joins, or when ACID transactions across multiple collections are frequently needed.
Use a time-series database (InfluxDB, TimescaleDB) for sensor data or metrics - MongoDB's document model is inefficient for high-frequency time-series writes.

Reference Files

https://www.mongodb.com/docs/manual/
https://mongoosejs.com/docs/guide.html