Serverless is awesome — but performance matters. Cold starts, heavy bundles, or noisy logs can hurt your Lambda’s speed and cost. Today, we’ll explore practical ways to improve Node.js performance in AWS Lambda using AWS SAM.

Whether you're building APIs or event-driven apps, these tips will help you write faster, leaner, and more cost-effective serverless functions.

🚀 1. Initialize Services Outside the Handler

This is the golden rule. Anything that can be reused between invocations should be initialized outside the handler.

// Good ✅
import { DynamoDBClient } from "@aws-sdk/client-dynamodb";

const dbClient = new DynamoDBClient({});

export const handler = async (event) => {
  // Use dbClient
};

Why? Because AWS reuses Lambda containers during warm starts, and re-initializing things like database clients inside the handler adds unnecessary overhead.

🧠 2. Use the Latest Node.js Runtime

Always use the latest supported Node.js version in AWS Lambda (at the time of writing: nodejs22.x).

Benefits:

  • Faster startup time
  • Better performance and memory handling
  • Access to newer language features 
Runtime: nodejs22.x

💪 3. Choose ARM Architecture (Graviton)

Graviton-based Lambdas (arm64) offer up to 34% better performance at 20% lower cost compared to x86_64.

In your template.yml

Properties:
  Architectures:
    - arm64

Unless you have specific native dependencies that don't support ARM, this is a smart switch.

📦 4. Keep Lambda Package Size Small

Smaller packages = faster cold starts. Here's how to shrink things down:

  • Use esbuild with minification and tree-shaking
  • Split shared code into Layers
  • Only include what you need — use External in your template.yml to exclude AWS SDK packages (they're included in the Lambda runtime) 
Metadata:
  BuildMethod: esbuild
  BuildProperties:
    Minify: true
    Target: "es2020"
    External:
      - "@aws-sdk/*"

🚫 5. Avoid Wildcard Imports

Avoid doing this:

// Bad ❌
import * as AWS from "aws-sdk";

Use specific imports instead:

// Good ✅
import { S3Client } from "@aws-sdk/client-s3";

📊 6. Adjust Your Lambda Memory

Memory = power in AWS Lambda. Increasing memory also gives you more CPU.

Run performance tests at different memory levels to find the sweet spot. Sometimes doubling memory cuts execution time in half — and reduces cost.

In your template.yml:

MemorySize: 1024

🧹 7. Don't Log Everything

Logging too much (especially in loops or high-frequency Lambdas) adds latency and cost.

Guidelines:

  • Use debug level only when actively troubleshooting
  • Keep info logs minimal
  • Reserve error logs for actual failures

If you're using structured loggers like Powertools Logger, you can control verbosity via environment variables.

const logger = new Logger({
  logLevel: process.env.LOG_LEVEL || "INFO",
});

Conclusion

Performance in Lambda isn’t just about speed — it’s about cost, cold start time, and developer experience.

✅ Recap:

  • Move services outside your handler
  • Use the latest Node.js runtime
  • Choose ARM64 for better performance
  • Keep packages small and clean
  • Avoid wildcard imports
  • Tune memory settings
  • Log wisely

With these techniques, your Node.js Lambdas will be lean, fast, and production-ready.