AWS for Backend Developers: Essential Services and Architecture Patterns

AWS has a few hundred services, and you need maybe a dozen of them to actually run a backend in production. This is the short list, based on what I've used, not the full catalog.

The Core Architecture

A production-ready AWS architecture for a Node.js API looks like this:

Internet
   │
   ▼
Route 53 (DNS)
   │
   ▼
CloudFront (CDN)
   │
   ▼
ALB (Application Load Balancer)
   │
   ├──▶ EC2 Instance 1 (Node.js API)
   ├──▶ EC2 Instance 2 (Node.js API)
   └──▶ EC2 Instance 3 (Node.js API)
          │
          ├──▶ RDS (PostgreSQL / MongoDB Atlas)
          ├──▶ ElastiCache (Redis)
          └──▶ S3 (File Storage)

EC2: Your Application Server

EC2 (Elastic Compute Cloud) provides virtual machines. For a Node.js API:

• Instance type: t3.medium (2 vCPUs, 4 GB RAM) is a solid starting point.

• AMI: Amazon Linux 2023 or Ubuntu 22.04.

• Security Group: Allow inbound on port 3000 from the ALB only, SSH on port 22 from your IP only.

Basic setup script (User Data):

#!/bin/bash
# Install Node.js 20
curl -fsSL https://rpm.nodesource.com/setup_20.x | bash -
yum install -y nodejs git

# Clone and start application
cd /home/ec2-user
git clone https://github.com/your-org/your-api.git
cd your-api
npm ci --only=production
npm run build

# Use PM2 for process management
npm install -g pm2
pm2 start dist/index.js --name api -i max
pm2 save
pm2 startup

Application Load Balancer (ALB)

The ALB distributes traffic across EC2 instances and provides:

• Health checks — automatically removes unhealthy instances.

• SSL termination — handles HTTPS so your Node.js app only deals with HTTP.

• Path-based routing — route /api/* to backend, /* to frontend.

Key configuration:

Listener: HTTPS :443
  Rule 1: Path /api/* → Target Group: backend-tg
  Rule 2: Default    → Target Group: frontend-tg

Target Group: backend-tg
  Health Check Path: /api/health
  Health Check Interval: 30s
  Healthy Threshold: 2
  Unhealthy Threshold: 3

Auto Scaling Group (ASG)

Auto Scaling automatically adjusts the number of EC2 instances based on demand:

Auto Scaling Group:
  Min Instances: 2
  Max Instances: 10
  Desired Capacity: 3

Scaling Policy (Target Tracking):
  Metric: Average CPU Utilization
  Target: 60%
  Scale Out Cooldown: 300s
  Scale In Cooldown: 300s

When CPU utilization exceeds 60 %, ASG launches new instances. When it drops, instances are terminated. You pay only for what you use.

RDS: Managed Databases

Amazon RDS handles database administration — backups, patching, replication, and failover. For PostgreSQL:

Instance: db.t3.medium
Storage: 100 GB gp3
Multi-AZ: Enabled (automatic failover)
Automated Backups: 7-day retention

Connection from Node.js:

import { Pool } from 'pg';

const pool = new Pool({
  connectionString: process.env.DATABASE_URL,
  max: 20,
  idleTimeoutMillis: 30000,
  connectionTimeoutMillis: 5000,
  ssl: { rejectUnauthorized: false },
});

S3: Object Storage

S3 stores files (user uploads, backups, static assets) with 99.999999999 % durability:

import { S3Client, PutObjectCommand } from '@aws-sdk/client-s3';

const s3 = new S3Client({ region: 'ap-south-1' });

async function uploadFile(key: string, body: Buffer, contentType: string) {
  await s3.send(
    new PutObjectCommand({
      Bucket: process.env.S3_BUCKET!,
      Key: key,
      Body: body,
      ContentType: contentType,
    })
  );
  return `https://${process.env.S3_BUCKET}.s3.amazonaws.com/${key}`;
}

Combine S3 with CloudFront for global CDN delivery of uploaded images and files.

SQS: Message Queues

Amazon SQS is a fully managed message queue. Use it for background job processing:

import { SQSClient, SendMessageCommand, ReceiveMessageCommand } from '@aws-sdk/client-sqs';

const sqs = new SQSClient({ region: 'ap-south-1' });
const QUEUE_URL = process.env.SQS_QUEUE_URL!;

// Producer
async function enqueueJob(payload: object) {
  await sqs.send(
    new SendMessageCommand({
      QueueUrl: QUEUE_URL,
      MessageBody: JSON.stringify(payload),
    })
  );
}

// Consumer (runs on a separate worker instance)
async function pollJobs() {
  const response = await sqs.send(
    new ReceiveMessageCommand({
      QueueUrl: QUEUE_URL,
      MaxNumberOfMessages: 10,
      WaitTimeSeconds: 20, // long polling
    })
  );
  return response.Messages || [];
}

ElastiCache: Managed Redis

ElastiCache provides managed Redis or Memcached. For session caching and API response caching:

Node Type: cache.t3.medium
Cluster Mode: Disabled (single-shard is enough for most APIs)
Multi-AZ: Enabled
Automatic Failover: Enabled

Your application connects to ElastiCache the same way it connects to local Redis — just update the REDIS_URL environment variable.

Cost Optimization Tips

• Use Reserved Instances for predictable workloads (save up to 60 %).

• Use Spot Instances for batch processing and non-critical workers.

• Enable S3 lifecycle policies to transition old objects to Glacier.

• Set up CloudWatch billing alarms to avoid surprise bills.

• Use t3 instances with burstable performance for development environments.

Wrapping Up

EC2, ALB, ASG, RDS, S3, SQS, and ElastiCache cover most of what a backend actually needs from AWS. Set a billing alarm before you need one — cloud costs have a way of growing quietly until they don't.