Stop 08 · Serverless & Event-Driven ArchitectureTopics · 1/9

Serverless Fundamentals

In short

Your college fest site gets zero visitors for eleven months and a flood for one week. Paying for a server all year for that week feels silly. Serverless bills only the flood.

Running code without provisioning or managing any servers directly — the platform handles all scaling, patching and availability, and you pay only for actual execution time rather than idle capacity.

The blob character hands shirts to a roadside ironing man at his wooden cart.

Think of it like this

The press-wala on the corner

You don't buy an iron, an ironing board and a coal stove. You hand three shirts to the press-wala, pay per piece, and collect them pressed. Exam week, ten shirts? He handles it. Serverless is paying per piece for compute.

Watch it happen

Requests per hour over one day

480

1/5Each cell is one 3-hour slot of the day. Overnight there are zero requests, so zero function copies run and you pay nothing.

The big ideas

What you need to know

Servers still exist

You just never see, patch or size them. The provider runs and scales the machines.

Event triggered

Code runs because something happened: an HTTP request, a file upload, a message, a schedule.

Pay per use

Billing is per request and per unit of run time and memory, not per hour of a running machine.

More than functions

Serverless also covers managed queues, databases, storage and API gateways you never provision.

See it in code

Walk through the code

yaml

1AWSTemplateFormatVersion: "2010-09-09"2Transform: AWS::Serverless-2016-10-313Resources:4  HelloFn:5    Type: AWS::Serverless::Function6    Properties:7      Runtime: nodejs22.x8      Handler: index.handler9      CodeUri: src/10      MemorySize: 25611      Timeout: 1012      Events:13        Hello:14          Type: HttpApi15          Properties: { Path: /hello, Method: get }

1/4The Transform line lets you use SAM's short serverless resource types.

An AWS SAM template. Run sam build and sam deploy --guided, and you get a public URL with no server anywhere in your config.

Save yourself the pain

Common mistakes

Oops

Thinking serverless means there are no limits.

Do this instead

Functions have time limits, memory caps and concurrency quotas. Read them before you design around them.

Oops

Assuming serverless is always cheaper.

Do this instead

It's cheapest for spiky or low traffic. Steady heavy load can cost more than a right-sized server.

Your mission

Hello, serverless

~35 min

Quick check

Which workload benefits most from serverless pricing?

Go deeper

1Learn the core trade-off: no server management, in exchange for execution time limits and less control.
2Identify a workload from earlier phases that would fit serverless well, and explain why.
3Understand serverless doesn't mean 'no servers exist' — it means you don't manage them.