Thread by @piotr_minkowski

💡No matter which Open JDK vendor you choose? ☕️🤔

Let's compare Oracle vs Eclipse Temurin. For my app (Spring Boot 3 + Mongo) diff is pretty significant:

Oracle: startup 12-13s, memory ~270M, 520req/s
Temurin: startup 8-9s, memory ~220M, 620req/s

See how I'm testing it 👇

I have a simple Spring Boot 3 🍃 app that connects to Mongo. Both app and db are running on Kubernetes. In my test I'm simulating 10 virtual users that continously send requests to the app (app inserts data to Mongo). Here's my app repo 👇
github.com/piomin/sample-…

I'm building the image from the source with Paketo Buildpacks (paketo.io/docs/howto/jav…) and Skaffold. Thanks to that I can simple switch between the vendors just by replacing buildpack in `skaffold.yaml`. Then I'm just running skaffold dev👇

(1/2) For load testing I'm using k6 (k6.io). I'm simulating 10 users sending reqs continuously for 60 seconds. Here's the result for Oracle OpenJDK (31k reqs -> 520r/s) 👇

(2/2) I'm simulating 10 users sending reqs continuously for 60 seconds also for Temurin. Here's the result for (37k reqs -> 620r/s) 👇

You can try it by yourself. Just clone the repo I provided in this thread. Then go to the `k8s` directory. The test is written in js (`load-tests.js` file). After deploying the app just run it as shown below. I'm curious what will be your result.

To read more about available JDK vendors and recommendations go to the site whichjdk.com