Closed codefromthecrypt closed 7 years ago
I've been meaning to get in touch since your talk at OSCON Austin, I was impressed with Zipkin. I'm the founder of Katacoda, we're an interactive technical learning platform for software developers. We allow users to access production-like environments directly in their browser for learning and experimenting.
About the issue, would it be useful to have an interactive scenario on how Prometheus and Zipkin work together? The scenario could discuss configuring Prometheus and how Zipkin metrics can be pushed into it. It would be able to highlight the benefits and how it can be configured with a real-system.
Here are two scenarios we collaborated on with other open source projects to highlight their initial use-case. More coming soon: https://katacoda.com/courses/coredns https://katacoda.com/courses/traefik
Would that help solve the problem? Long term, I thought additional scenarios could cover ES, MySQL etc
Thanks for reaching out, Ben. This sounds like a neat way to learn. cc @abesto
Huh. Agree, this looks interesting. I've recently played around with Prometheus, running it locally with a simple setup was trivial, so I think we could add it to a docker-compose.yml
with minimal fuss. Adding this to my todo list, no commitment for now.
@BenHall, given that Zipkin already has support for exposing a Prometheus-compatible endpoint, I think a Zipkin-Prometheus specific scenario would provide significantly less value than a "Prometheus intro" scenario. Of course it'd make total sense to then build a Prometheus-Zipkin scenario on top of it.
(on the training aside) actually agree w/ @abesto in hind-sight. we probably would get more bang for the buck adding to prometheus, or directing towards a generic zipkin one
To show how server metrics can be useful, it would be best to setup prometheus (or some other setup) accordingly. Otherwise, they can be a bit abstract.
cc @kristofa @klette @abesto @klingerf @gena01 for ideas, too