Looking to the future
Viktor Farcic: So, what's next, then?
Andy Clemenko: In the near future, I see serverless picking up some momentum, but I'm still waiting for serverless to be actually written into the lower-level orchestrator directly, and not as it currently is, which is as an extra layer on top. To me, serverless is just a rapid reaction scheduler, to some extent.
Elias Pereira has done some really awesome stuff with OpenVAS, to the point where it's got self-autoscaling of containers because it's deploying its own Prometheus. To me, conceptually, having similar functions at multiple layers seems redundant. So, let me ask this: if we can take OpenVAS and build it into the lower orchestrator, why don't we build into right into Kube or right into Swarm?
At least that way I'm advocating for a 38 top-level object. But the idea, though, is that if you have more batch processes like serverless, they can still use the same schedule. You don't need to build...