Serverless

Serverless computing is a cloud execution model where the cloud provider manages the infrastructure, automatically provisions resources, and scales based on demand. You write code; the platform handles everything else.

Key Characteristics

• No server management: No provisioning, patching, or scaling decisions
• Pay-per-use: Charged only for actual execution time
• Auto-scaling: Scales from zero to thousands of instances automatically
• Event-driven: Functions triggered by events (HTTP, queues, schedules)

Serverless vs Traditional

Aspect	Serverless	Traditional Server
Scaling	Automatic	Manual/configured
Pricing	Per execution	Per hour/month
Cold starts	Yes	No
Maintenance	Provider	You
Control	Limited	Full
State	Stateless	Stateful possible

AWS Serverless Stack

Service	Purpose
Lambda	Compute (functions)
API Gateway	HTTP APIs
DynamoDB	Database
S3	File storage
SQS / SNS	Messaging
EventBridge	Event routing
Cognito	Authentication
CloudFront	CDN

Common Patterns

API Backend

CloudFront → API Gateway → Lambda → DynamoDB

Event Processing

S3 Upload → Lambda → Process → Store Result

Scheduled Tasks

EventBridge (cron) → Lambda → Task Execution

Queue Processing

SQS → Lambda → Process Message → External Service

What We Like

• Operational simplicity: Focus on code, not infrastructure
• Cost efficiency: No idle costs, great for variable workloads
• Scalability: Handle traffic spikes without planning
• Speed to deploy: From code to production in minutes
• Reduced risk: Less infrastructure to secure

What We Don't Like

• Cold starts: Initial latency on first invocation
• Vendor lock-in: Architectures tied to specific providers
• Debugging: Distributed systems are harder to trace
• Statelessness: Requires rethinking data patterns
• Execution limits: Timeouts, memory caps, package sizes

Cold Start Mitigation

• Provisioned Concurrency: Keep instances warm (costs money)
• Smaller packages: Faster initialisation
• Choose runtime wisely: Node.js/Python start faster than Java
• Keep functions warm: Scheduled pings (hacky but works)

When to Use Serverless

Good fit:

• Variable or unpredictable traffic
• Event-driven workloads
• APIs with sporadic usage
• Background processing
• Startups and MVPs

Consider alternatives for:

• Consistent high traffic (EC2 may be cheaper)
• Long-running processes
• Stateful applications
• Ultra-low latency requirements

Beyond AWS

Provider	Service
Google Cloud	Cloud Functions, Cloud Run
Azure	Azure Functions
Cloudflare	Workers
Vercel	Edge Functions