Skip to content

mbwilding/lambda-benchmark

Repository files navigation

AWS Lambda Performance Benchmark

Benchmark AWS Lambda performance across various runtimes, architectures, and memory configurations.

Overview

This benchmarking tool performs the following operations:

  1. Writes the current iteration value to an S3 key, 500 times.
  2. Deletes the S3 key after all 500 writes are completed.
  3. Repeats steps 1 and 2, 10 times for each runtime/architecture/memory configuration.
  4. Calculates the average time taken for these operations and presents the results.

Note: Everything is deployed using CloudFormation. Deleting the stack will remove all associated resources, ensuring there's no residual footprint. No resources are created dynamically.

Configuration

To adjust the settings, edit the parameters.yml file and redeploy Stack via GitHub Actions.

Results

See the benchmark results without any local setup. Click the link below for interactive graphs:

View Benchmark Graphs

Extending Benchmarks with a New Runtime

To add a new runtime:

  1. Create a new directory under runtimes/ named after the runtime. For instance: runtimes/rust/.
  2. In this directory, include your implementation. Also, add a manifest.yml and build.sh.
  3. Redeploy Runtimes and then Stack via GitHub Actions.