What Is AWS Lambda? Serverless Functions Explained

AWS Lambda is a serverless compute service: you upload a function, AWS runs it on demand in response to events, and you pay only for the milliseconds it executes.

Lambda lets you run code without provisioning or managing servers. You package a function, define what triggers it, and AWS handles scaling, patching, and availability. It is the canonical example of "serverless" compute: there are still servers, but you never see or manage them.

How a Lambda function runs

You deploy a handler function in a supported runtime (Node.js, Python, Java, Go, and others) or as a container image. When a trigger fires, Lambda spins up an execution environment, runs your handler, and freezes or tears it down afterward. Concurrency scales automatically as events arrive.

What triggers a function

HTTP requests via API Gateway or a function URL.
Events from S3, DynamoDB streams, or SQS and SNS.
Scheduled invocations on a cron-like timer.
Direct calls from other services or the AWS SDK.

Cold starts and limits

The first invocation of an idle function pays a cold start while the environment initializes. Functions have a configurable memory size, a maximum timeout, and a deployment package size limit. Keeping packages small and dependencies lean reduces cold-start latency.

Pricing model

You pay per request plus compute time measured in gigabyte-milliseconds, so a function that rarely runs costs almost nothing. This makes Lambda attractive for spiky or event-driven workloads where a constantly running server would be wasteful.

Role in CI/CD

A deploy job zips your function code (or builds a container image), then updates the function with the AWS CLI, SAM, the Serverless Framework, or Terraform. The pipeline authenticates with an IAM role, publishes a new version, and can shift traffic gradually using aliases for safer rollouts. Tests and linting run earlier in CI before the deploy step ever fires.

Key takeaways

Lambda runs code on demand in response to events, with no servers to manage.
You pay per request and per millisecond of execution, ideal for spiky workloads.
Pipelines deploy functions by updating code or images via IAM-authenticated tooling.