This spike explores using AWS LLRT as a faster JavaScript runtime for Lambda functions.

It pairs a working repository with a writeup so the runtime tradeoffs are easier to test and understand.

The implementation details below are synced from the project README.

This project demonstrates how to build a serverless API using AWS Lambda with LLRT (Low Latency Runtime) - a lightweight JavaScript runtime optimized for AWS Lambda.

The function are behind simple HTTP API Gateway endpoints.

Features

• Fast cold starts with LLRT runtime
• Multiple API endpoints (/hello and /goodbye)
• ARM64 architecture for better performance/cost ratio
• Custom Serverless Framework plugin for deployment validation
• Automated LLRT bootstrap setup
• 🔧 LLRT Compatibility Checker - Verify code compatibility before deployment

Prerequisites

• Node.js 18 or later
• AWS CLI configured with appropriate credentials
• Serverless Framework CLI (npm install -g serverless)

Quick Start

1. Clone the repository
2. Install dependencies:

npm install

3. Set up LLRT bootstrap:

npm run setup

4. Test LLRT compatibility (optional):

npm run check:hello
npm run check:goodbye
# Or run all tests
npm test

5. Deploy to AWS:

npm run deploy

API Endpoints

Hello Endpoint

• URL: GET /hello
• Query Parameters:
  • name (optional) - Name to greet
• Example Response:

{
  "message": "Hello World from LLRT!",
  "timestamp": "2024-03-14T12:00:00.000Z",
  "path": "/hello",
  "method": "GET"
}

Goodbye Endpoint

• URL: GET /goodbye
• Query Parameters:
  • name (optional) - Name to bid farewell
• Example Response:

{
  "message": "Goodbye World, thanks for using LLRT!",
  "timestamp": "2024-03-14T12:00:00.000Z",
  "path": "/goodbye",
  "method": "GET"
}

Project Structure

.
├── bootstrap # LLRT runtime
├── bin/ # CLI tools
│   └── llrt-check.js # LLRT compatibility checker
├── docs/ # Documentation
│   └── llrt-check.md # Checker documentation
├── node_modules # Node.js dependencies
├── package.json
├── serverless.yml
├── src/ # Source code
│   ├── hello.js
│   ├── goodbye.js
│   └── test-function.js
├── scripts/ # Setup and utility scripts
│   ├── setup.js
│   ├── get-compat.js
│   └── test-checker.js
└── llrt-compatibility.json # Compatibility matrix

Deployment

To deploy the service, run:

npm run deploy

Cleanup

To delete the service, run:

npm run remove

LLRT Compatibility Checker

This project includes a custom compatibility checker to verify your code will work with LLRT before deployment:

# Check specific functions
npm run check:hello
npm run check:goodbye

# Check any JavaScript file
./bin/llrt-check.js path/to/your/function.js

# Run all compatibility tests
npm test

The checker performs:

• Static Analysis: Scans for unsupported modules and features
• Runtime Testing: Executes code in both Node.js and LLRT environments
• Performance Comparison: Reports execution time differences

See docs/llrt-check.md for detailed usage instructions.

Notes

• You can deploy LLRT in many ways, see https://github.com/awslabs/llrt?tab=readme-ov-file#configure-lambda-functions-to-use-llrt
• See Serverless Framework docs for more information on how to deploy and manage your service.

Performance results

You can perf test with node perf-test.js

Average response time: 54.05ms. Not too shabby.

=== PERFORMANCE COMPARISON SUMMARY ===
Comparing Node.js vs LLRT implementations:

Hello Function:
  Average: 149.96ms (Node) vs 32.37ms (LLRT) - 78.4% faster
  Min: 127.19ms (Node) vs 22.47ms (LLRT) - 82.3% faster
  Max: 182.75ms (Node) vs 51.00ms (LLRT) - 72.1% faster

Goodbye Function:
  Average: 133.00ms (Node) vs 30.40ms (LLRT) - 77.1% faster
  Min: 119.05ms (Node) vs 22.92ms (LLRT) - 80.7% faster
  Max: 148.67ms (Node) vs 48.12ms (LLRT) - 67.6% faster

Test Function:
  Average: 135.46ms (Node) vs 30.98ms (LLRT) - 77.1% faster
  Min: 124.46ms (Node) vs 21.34ms (LLRT) - 82.9% faster
  Max: 151.80ms (Node) vs 49.26ms (LLRT) - 67.6% faster

=== OVERALL SUMMARY ===
LLRT is on average 77.6% faster than Node.js
Average response times: 139.47ms (Node) vs 31.25ms (LLRT)