serverless-projects / lambda-memory-performance-benchmark

Performance and cost benchmark tool for AWS Lambda on memory sizes 📈⏱

AWS Lambda Memory Performance Benchmark

Motivation

• Understand how RAM selection affect Lambda's performance (on Medium).
• Code a tool that measures Lambda function performance (duration and cost) with several memory sizes.

Setup

git clone [e-mail]:epsagon/lambda-memory-performance-benchmark.git
cd lambda-memory-performance-benchmark/
pip install -r requirements.txt
python benchmark.py -f <function_name> -r <function_region> -p <payload_filename>

Usage

Basic run:

python benchmark.py -f lambda-performance-benchmark -r us-east-1 -p fibonacci-function/payload.json

Fibonacci's Last Result (February 9th, 2018)

Chart:

Table:

Memory Size	Duration (in ms)	Price Per 1M Invocations (in $)
128MB	376.05	83.2
192MB	250.24	93.9
256MB	198.56	83.4
512MB	98.36	83.4
768MB	65.48	125.0
1024MB	45.31	166.7
1536MB	30.53	250.1
2048MB	25.21	333.4
2560MB	25.36	416.8
3008MB	25.15	489.7