Functions and Event-Driven Processing: Going Serverless

Core Concept

• FaaS (Functions-as-a-Service) = event-driven, short-lived compute
• Functions:
  • Run on-demand
  • Are stateless
  • Triggered by events or requests

Often called serverless (but not exactly the same)

When to Use FaaS

Best for:

• Event-driven tasks
• Short-lived processing
• Low/variable traffic

Examples:

• Sending emails
• Notifications
• API transformations

Benefits

1. Extremely simple deployment

• Just write code → no infra management
• Faster time to production

2. Auto-scaling

• Automatically scales with traffic
• Handles failures automatically

3. Strong decoupling

• Each function is independent
• Encourages modular systems

4. Cost efficient (for low traffic)

• Pay per request
• No idle cost

Challenges

1. Hard to debug & observe

• No clear system view
• Complex interactions between functions

2. No local state

• Must use external storage
• Adds latency + complexity

3. Risk of hidden loops

• Functions calling each other → infinite loops
• Hard to detect due to decoupling

4. Monitoring is critical

• Need strong:
  • Logging
  • Alerting
  • Observability

When NOT to Use FaaS

1. Long-running tasks

• Example:
  • Video processing
  • Batch jobs

FaaS has execution time limits

2. Memory-heavy workloads

• Cold start + data loading → high latency

3. High sustained traffic

• Pay-per-request becomes expensive
• Containers/VMs become cheaper

Key Patterns

1. Decorator Pattern (very important)

• Function sits between client and service
• Transforms request/response

Example:

• Add default values
• Validate input

Lightweight and stateless → perfect for FaaS

2. Event Handling

• Trigger functions on events:
  • User signup
  • File upload
  • Login

Example:

• 2FA → send OTP asynchronously

Improves UX by offloading slow tasks

3. Event Pipelines

• Chain multiple functions:
  • Each does one small task
• Connected via events/webhooks

Example:

Signup → Send email → Subscribe user → Analytics

Like microservices but event-driven

FaaS vs Microservices

Key Insights

• FaaS = extreme decoupling
• Great for:
  • Glue logic
  • Async tasks
• Not for:
  • Core heavy systems

Trade-offs

Pros

• Fast development
• Auto scaling
• Cost-efficient (low usage)
• Highly modular

Cons

• Hard debugging
• Cold starts
• Costly at scale
• Limited execution model

One-line Summary

FaaS enables event-driven, stateless, auto-scaling functions that are ideal for lightweight, asynchronous tasks—but not suitable for long-running, stateful, or high-throughput systems.