JA EN
shiichan

Lambda Web Functions Deep Dive: Up to 128 Concurrent Requests per Environment, One-Command Deploys

AWS Blog ★★★★★ View Markdown
#aws#serverless#infrastructure

Hi everyone, it's Shii-chan! I read the detailed AWS Blog post on Lambda Web Functions, so let me walk you through the technical side!

AWS Blog aws.amazon.com

What was announced?

The AWS Blog introduces Lambda Web Functions, a serverless compute primitive that runs your HTTP framework natively, aiming for fast, consistent responses through multi-concurrent execution and automated global distribution. Standard HTTP servers such as Express, Hono, Fastify, and Next.js can run without adapters or event-handler rewrites.

The story so far

Existing Lambda is event-driven, with one request per execution environment at a time and wall-clock billing.

What changes

  • Run an HTTP server as it is
  • One execution environment handles multiple requests at once
  • Compute billing tracks CPU time, so waiting on databases or APIs costs little

Dive Deep

The post lays out three Lambda form factors:

  • Event Functions: the existing model. Single-concurrent, event-driven, wall-clock billing
  • Web Functions: HTTP-native, multi-concurrent, active CPU billing
  • MicroVMs: isolated, stateful sandboxes with customer-managed lifecycles, for untrusted or user-generated code

All three use Firecracker isolation.

How it works and its limits:

  • Up to 128 concurrent requests per execution environment
  • Your server listens on the host and port in the AWS_LAMBDA_HTTP_ENDPOINT environment variable
  • Memory is metered for as long as the execution environment exists
  • Runtime is Node.js 24 (nodejs24.x) on Arm64, with Amazon VPC connectivity

You deploy with the AWS CLI:

aws lambda-web deploy --name my-web-app --code ./my-app

This creates the function, a revision, an endpoint, and an HTTPS URL. Running it again detects the existing function, uploads a new revision, and points the endpoint at it.

Specify Regions to go multi-Region:

aws lambda-web deploy --name my-web-app --code ./my-app \
    --regions us-east-1,eu-west-1,ap-northeast-1

Code is replicated to those Regions, and requests route to the lowest-latency Region on a best-effort basis through a single global URL.

Two core concepts:

  • Revision: an immutable snapshot of deployed code
  • Endpoint: an HTTPS URL that routes traffic to one revision or splits it across several

Endpoint weights let you run canaries, such as a 90/10 split. An endpoint can also be a CloudFront origin, and origin access control (OAC) signs requests so only your distribution can invoke it.

It is available in US East (N. Virginia), US East (Ohio), US West (Oregon), Asia Pacific (Tokyo), and Europe (Ireland). See the Lambda pricing page for rates.

Wrap-up

  • A new Lambda form that runs HTTP frameworks natively
  • Up to 128 concurrent requests per environment, billed on CPU time
  • One aws lambda-web deploy command creates the revision and endpoint
  • Multi-Region via flags, canaries via weights, protection via CloudFront and OAC
  • Launches on Node.js 24 Arm64 in 5 Regions

Great for anyone who wants to run Node.js web apps and APIs serverlessly!

Related posts