Python vs. Node.js (AWS Lambda) performance benchmark (2023)

▬▬▬▬▬ Experience & Location 💼 ▬▬▬▬▬
► I’m a Senior Software Engineer at Juniper Networks (12+ years of experience)
► Located in San Francisco Bay Area, CA (US citizen)

▬▬▬▬▬▬ Connect with me 👋 ▬▬▬▬▬▬

▬▬▬▬▬▬ Related videos 👨‍🏫 ▬▬▬▬▬▬

=========
⏱️TIMESTAMPS⏱️
0:00 Intro
0:37 Create Python Lambda Function
1:40 Terraform Code to Reproduce
2:07 Performance Benchmark

=========
Source Code

#AWS #Lambda #DevOps

[Meamka]

Seems like the test is heavily depend on s3 and dynamo performance. When you try to benchmark language maybe it is better to use only language without any external connections and dependencies.

[bryanmonsalvatge]

Thanks for the comparisons. Would love to see a Go vs Python vs Rust comparison on Lambda since Go and Python seem most performant so far. And the Rust SDK is starting to become more mature.

[TheSpeakingClub]

Hi Anton! Thank you for the great lessons!

[spellskreenful]

Looks like time to compile node JS is longer than time of interpretation of Python, if service could get compiled code, node will be way faster

[agatehao]

Hmm... the result looks very strange to me. I can't believe that python is faster than node. For my experience, node should be way faster than python... This makes me curious how lambda execute those applications.

[NiklasHeer]

I was surprise to see the results of my silly project (speed-comparison) in this video. (0:22)
I've worked on the repo and there were numerous improvements made, so these results shown here are out of date.

[stxnw]

javascript devs try not to be upset about python being faster (IMPOSSIBLE)

[erbhargavvyas]

Awesome meterial.
I think one thing can be changed for all your test (i.e. go, python, rust, node) the test tool.
As I understand the test tool is of javascript. Maybe change it. A test tool written in go or C/C++ can produce different results 'cause the latency of the tool itself might have come into account.

Please read this comment for all 3 videos.

[fiords]

Wow, unexpected results. Seems Python is run with pypy or smth faster?

[techmech]

Well what are you trying to prove? Node js is never good for CPU intensive task. You need to create a bigger test case with multiple data intensive task and check. Also we need to refactor understand lamba cold starting time. Thanks!

[gshan]

I use plantuml diagram fr serverless arch diagram
Which one u use?

[AndrewTSq]

either the python library is C.. or the nodejs is the worst optimised javascript you ever seen

[httuksu]

Bad test as the tested languages do almost nothing.