Skip to main content

Lesson 6: What Can You Build with Serverless?

Applicability Matrix: Serverless Use Cases​

Serverless works best for event-driven, bursty, and decoupled workloads. Use it when you want fast iteration, pay-per-use economics, and independent scaling.

Simple Explanation​

What it is​

This lesson is about deciding fit. Some workloads are perfect for serverless, some are borderline, and some should avoid it entirely.

Why we need it​

Choosing the wrong compute model can burn budget and time. A quick fit check prevents painful rewrites later.

Benefits​

  • Clear decision rules that help teams align quickly.
  • Faster planning because you know what to prototype first.
  • Lower risk when you avoid bad-fit workloads early.

Tradeoffs​

  • Borderline cases need testing and measurement.
  • Fit depends on traffic and latency, not just the type of app.

Real-world examples (architecture only)​

  • Good fit: API for mobile app → Function → NoSQL database.
  • Mixed fit: Streaming analytics → Function fan-out → Storage.
  • Poor fit: Always-on low-latency trading engine → Containers or VMs.

Key Topics (Outline)​

  • Web APIs and microservices
  • Real-time data processing (streaming)
  • Scheduled jobs and batch processing
  • Image/video processing pipelines
  • IoT backends
  • Machine learning inference
  • Content delivery and static sites
  • ChatOps and automation
  • Mobile app backends
  • Data ETL pipelines

Quick Heuristics​

  • Good fit: HTTP APIs, background processing, file/event processing, automation
  • Mixed fit: Streaming and batch jobs (depends on duration and throughput)
  • Poor fit: Always-on, low-latency, long-running workloads

Project (Cloud-Agnostic)​

Pick one workload from the list and justify whether serverless is a good fit.

Deliverables:

  1. Describe the workload and its traffic pattern.
  2. Explain whether serverless fits and why.
  3. Map the architecture to AWS or GCP services.

If you want feedback, email your write-up to maarifaarchitect@gmail.com.

References​