MongoDB in 2026: Still the Right Choice for Many Applications

MongoDB 8.0 delivers major query performance improvements (up to 30% faster reads on large collections) and Atlas Vector Search is now first-class — making it a natural fit for AI-augmented applications. Mongoose 8.x improves TypeScript support dramatically. Here's how to use both correctly.

Schema Design Principles

Embed vs Reference — The Golden Rule

• Embed when data is accessed together and the sub-document is bounded in size (comments under a post ≤ 100 items)
• Reference when data is large, grows unbounded, or is queried independently (users, products)

// ✅ Good — embed small, stable data
const PostSchema = new Schema({
  title: String,
  tags: [String],                 // embedded array, bounded
  author: { type: ObjectId, ref: "User" }, // reference — User queried independently
});

// ❌ Bad — embedding unbounded array
const UserSchema = new Schema({
  posts: [{ type: ObjectId, ref: "Post" }], // could grow to millions
});

Mongoose 8.x TypeScript Integration

import { Schema, model, InferSchemaType } from "mongoose";

const blogSchema = new Schema({
  title:     { type: String, required: true },
  slug:      { type: String, required: true, unique: true },
  status:    { type: String, enum: ["draft","published"], default: "draft" },
  publishedAt: Date,
}, { timestamps: true });

type Blog = InferSchemaType<typeof blogSchema>;
// Blog is fully typed — no manual interface needed

export const Blog = model("Blog", blogSchema);

Indexing Strategy

// Compound index — matches query pattern exactly
blogSchema.index({ status: 1, publishedAt: -1 });

// Text search
blogSchema.index({ title: "text", content: "text" });

// TTL index — auto-delete expired documents
tokenSchema.index({ expiresAt: 1 }, { expireAfterSeconds: 0 });

Explain Your Queries

const explain = await Blog
  .find({ status: "published" })
  .sort({ publishedAt: -1 })
  .explain("executionStats");

console.log(explain.executionStats.totalDocsExamined);
// Should equal totalDocsReturned — if not, you need an index

Connection Pooling in Serverless

// lib/db/connect.ts — singleton pattern for Next.js
let cached = global.mongoose;

export default async function dbConnect() {
  if (cached?.conn) return cached.conn;

  cached.promise = cached.promise || mongoose.connect(MONGODB_URI, {
    maxPoolSize: 10,
    serverSelectionTimeoutMS: 5000,
  });

  cached.conn = await cached.promise;
  return cached.conn;
}

Atlas Vector Search for AI Features

// Store embeddings alongside documents
const articleSchema = new Schema({
  content: String,
  embedding: [Number],  // 1536-dim for text-embedding-3-small
});

// Vector search pipeline
const results = await Article.aggregate([{
  $vectorSearch: {
    index: "default",
    queryVector: await embed(userQuery),
    path: "embedding",
    numCandidates: 100,
    limit: 5,
  }
}]);