How to build a CRUD API on Amazon DynamoDB

Amazon DynamoDB is a fully managed NoSQL key-value and document database with single-digit millisecond latency at any scale. Unlike relational databases, you design access patterns first, then a key schema that serves them. This tutorial builds a CRUD layer in Node.js with the AWS SDK v3.

Prerequisites

• An AWS account and Node.js 20.
• @aws-sdk/client-dynamodb and @aws-sdk/lib-dynamodb installed.

Steps

1. Design the key schema

For an orders service keyed by customer, use a partition key pk (CUSTOMER#123) and sort key sk (ORDER#2026-01-01). This groups a customer's orders together for efficient queries.

2. Create the table

aws dynamodb create-table --table-name orders \
  --attribute-definitions AttributeName=pk,AttributeType=S AttributeName=sk,AttributeType=S \
  --key-schema AttributeName=pk,KeyType=HASH AttributeName=sk,KeyType=RANGE \
  --billing-mode PAY_PER_REQUEST

On-demand billing avoids capacity planning.

3. Put and get items

import { DynamoDBClient } from "@aws-sdk/client-dynamodb";
import { DynamoDBDocumentClient, PutCommand, GetCommand } from "@aws-sdk/lib-dynamodb";
const doc = DynamoDBDocumentClient.from(new DynamoDBClient({}));

await doc.send(new PutCommand({ TableName: "orders", Item: { pk: "CUSTOMER#123", sk: "ORDER#1", total: 42 } }));
const { Item } = await doc.send(new GetCommand({ TableName: "orders", Key: { pk: "CUSTOMER#123", sk: "ORDER#1" } }));

4. Update items

import { UpdateCommand } from "@aws-sdk/lib-dynamodb";
await doc.send(new UpdateCommand({
  TableName: "orders", Key: { pk: "CUSTOMER#123", sk: "ORDER#1" },
  UpdateExpression: "SET #t = :t", ExpressionAttributeNames: { "#t": "total" }, ExpressionAttributeValues: { ":t": 50 }
}));

5. Delete items

import { DeleteCommand } from "@aws-sdk/lib-dynamodb";
await doc.send(new DeleteCommand({ TableName: "orders", Key: { pk: "CUSTOMER#123", sk: "ORDER#1" } }));

6. Query an index

Fetch all of a customer's orders with a Query on the partition key, or add a Global Secondary Index to query by another attribute such as status.

Verification

Run each operation and confirm the returned items. Use aws dynamodb scan --table-name orders during development to inspect contents (avoid scans in production). Confirm a query returns only the targeted partition.

Next Steps

Add optimistic locking with condition expressions, enable point-in-time recovery, and stream changes to Lambda with DynamoDB Streams.