otel-demo
OpenTelemetry auto-instrumentation demo. It features a microservices-based travel app and Jaeger running inside a Docker Compose environment.
Blog post: Auto-instrumentation with OpenTelemetry
Windows setup
Do the following from an elevated command shell:
- Install the Chocolatey package manager.
- Install OpenJDK:
choco install openjdk. - Install Make:
choco install make. - Install Docker Desktop for Windows:
choco install docker-desktop. There's some setup, too:- You'll need to decide whether you want to use WSL 2 (Linux) or else Hyper-V as a Docker backend. I recommend WSL 2, as it has better performance. Either way, you'll need to restart your computer.
- If you chose Hyper-V, go to Docker Desktop > Settings > Resources, and give yourself 8 GB of memory.
- Optional: If you want to play around with the app itself, install Visual Studio Code:
choco install vscode. Include Java support, and install the Docker and Lombok extensions too. The Docker extension will make it easier to view container logs and shell into containers.
Build
From the top of the project directory, build the app and the Docker images using a single command:
$ make
Run
To start up all the containers, it's just:
$ docker-compose up
Endpoints
- Sample app:
- UI: http://localhost:8080
- API: http://localhost:8081
- Provider 1:
- API: http://localhost:8082
- DB: mysql://localhost:3306
- Provider 2:
- API: http://localhost:8083
- DB: mysql://localhost:3307
- Jaeger UI: http://localhost:16686
To try it out, just hit the sample app UI a few times, and then look at the traces in Jaeger.
OpenTelemetry vs spring sleuth bench
This version has the UI and jaeger disabled to save RAM and avoid swapping
Choose to activate OTEL or SLEUTH in docker-compose.yml
docker-compose up
In another terminal launch the test, it lasts 1 minute:
jmeter -n -t perftest.jmx
Get requests per minute and avg rtt in ms:
wc -l out.jtl &&awk -F, '{ total += $2; count++ } END { print total/count }' out.jtl