Closed psschwei closed 1 year ago
A couple of questions here @psschwei , I'm not enough familiarized with Knative yet so sorry for my ignorance.
Which would be the difference between what we have right now and what you are proposing? My experience working with k8s
+ docker
right now is the next one:
dockerize
the APIauto-scaling
and so on. How the database is independent we can do that (I know that they are not independent right now but it is where we want to be in a short term).Where would you say in those steps would be the difference? Do you have some link/documentation that I could read to research a bit more? Thanks as always @psschwei 💯
Btw and just to confirm the context, is this issue related with some of these conversations that we had?
My main thought process here was :
That's about as far as I've gotten :smile:
I think this was partially motivated by some of the issues we saw in #255 and some of the Keycloak PRs. Knative could help here, but there would also be some additional overhead with the knative components... so really just an open question at this point.
Knative docs for Serving (which is the component we'd use) are here: https://knative.dev/docs/serving/
Oh, I see your point. I will read your documentation and provide my feedback!
The only question here is... Can we reuse it outside of kubernetes? We should be much agnostic as possible... it means, we need to have the ability to deploy our stack outside of kubernetes environment... but if we consider that Knative can be great for the kubernetes deployment or whatever, we can discuss as well... but to have in mind that we need to provide the option to deploy the stack in other places (not just kubernetes)
That's fair... I was assuming those parts would always run on kubernetes / openshift as part of the more general hybrid / multi cloud strategy
Currently, we run a couple of standalone services as part of quantum serverless: the gateway and the repository. As these are containerized services, it might be worthwhile to run them as Knative services, which would simplify the setup and provide some benefits (like autoscaling and networking) that we might want to take advantage of.
Opening this issue to explore this topic a bit...