Compute — EC2, Lambda & Choosing the Right Model

Compute — EC2, Lambda & Choosing the Right Model

Two Mental Models of Compute

EC2: I need a server. Give me a virtual machine, I'll install what I need and keep it running.

Lambda: I have a function. Run it whenever this event happens, for as long as it takes, and charge me only for that execution.

Every other AWS compute option (ECS, Fargate, App Runner) lives between these two extremes.

EC2 in Depth

EC2 instances are virtual machines running on AWS hardware. You choose:

Instance families:

• t (burstable) — cheapest, earns CPU credits when idle, spends them on spikes. Good for dev environments, low-traffic apps.
• m (general purpose) — balanced CPU/RAM. The default for most workloads.
• c (compute optimized) — high CPU-to-RAM ratio. Good for batch processing, video encoding.
• r (memory optimized) — high RAM. Good for in-memory caches, large databases.

Key EC2 concepts:

# Connect to an instance via SSH
ssh -i my-key.pem ec2-user@<public-ip>

# User data — bootstrap script that runs on first launch
#!/bin/bash
yum update -y
yum install -y nodejs
node -e "require('http').createServer((r,s)=>s.end('ok')).listen(3000)"

Security Groups are the primary network control for EC2:

Inbound rules:
  Port 22 (SSH)  — my IP only (0.0.0.0/0 is a security risk)
  Port 443 (HTTPS) — 0.0.0.0/0
  Port 80 (HTTP)   — 0.0.0.0/0

Outbound rules:
  All traffic — 0.0.0.0/0 (default, allows updates/API calls)

Lambda in Depth

Lambda functions are event-driven. AWS manages all infrastructure; you write the handler.

// A simple Lambda handler (Node.js 20.x runtime)
export const handler = async (event) => {
  console.log('Event:', JSON.stringify(event))

  return {
    statusCode: 200,
    body: JSON.stringify({ message: 'Hello from Lambda' }),
  }
}

Lambda limits to know:

• Max execution time: 15 minutes
• Memory: 128 MB – 10 GB (CPU scales with memory)
• Deployment package: 50 MB zipped (or 250 MB via Lambda Layers)
• Concurrent executions: 1,000 per region by default (soft limit, can increase)

Common Lambda triggers:

• API Gateway → HTTP request
• S3 → file uploaded/deleted
• SQS → message on a queue
• EventBridge → scheduled cron or custom event
• DynamoDB Streams → database change

The Decision Framework

Cost intuition:

• Lambda: ~$0.20 per 1M invocations + duration. Near-zero cost at low volume.
• t3.micro: ~$0.0104/hr = ~$7.50/month. Cheaper at sustained high request volume.
• The crossover is roughly 1–5M monthly invocations depending on duration.