I'm doing my school project, which requires me to build an AI management platform with requirements listed below:
- microservice architecture
- train and deploy AI in docker
- provide API to access deployed AI models
- AI models version management
- dataset management
- CI/CD
I'm trying to figure out a solution for backend.
Currently I select springcloud to provide backend service.
But I want to know that can I use springcloud to manage k8s? If I can, is there two k8s clusters, one for my springcloud project, one for my AI dockers?
Or maybe I should use k8s+istio+ingress+...... to realize this? If so, is there any documentations and examples I can learn from?
You should think about using some other technology for automatic pod creation based on certain conditions like apache airflow with python, which supports creating new pods using Kubernetes Pod Operator. Each operator can start your spring boot micro services build as docker image.
Example Flow: Airflow ---> Python ---> Kubernetes Pod Operator --> Manage Pods on k8s