> If you’re going with a microservice:
> A Kubernetes cluster
> A load balancer
> Multiple compute instances[...]
> Jaeger/Zipkin for distributed tracing
To be fair, using K8s + helm I was able to install logging, grafana, prometheus very easily. If you leverage on Helm3 and Bitnami [1] helm charts, you can go fast.
Also, you can use pipeline (like github/bitbucket pipelines) to deploy and remove jenkins completly: I have done it and it is a viable solution (although with some lock-in).
So the complexity is a bit less if you study enough well your setup, but you must take time to plan your solution.
After three years of K8s study, in my humble option K8s is far better compared to docker swarm, even for tiny projects, with k8s as a cloud managed solution (even small provider had it nowadays).
Also, you can use pipeline (like github/bitbucket pipelines) to deploy and remove jenkins completly: I have done it and it is a viable solution (although with some lock-in).
So the complexity is a bit less if you study enough well your setup, but you must take time to plan your solution.
After three years of K8s study, in my humble option K8s is far better compared to docker swarm, even for tiny projects, with k8s as a cloud managed solution (even small provider had it nowadays).
[1]: https://bitnami.com/stack/postgresql/helm