So adding to Brian’s point about the community aspect, I think that we’re trying to evolve our own products and make them more usable and offer things to help customers solve problems, and I think things like Virtual Kubelet definitely do that, but I think more importantly though is the advancement of the community and the technology, and we’re — you know, Kubernetes is still so new when we’re trying to figure out innovative ways to use it and run it in different scenarios, for different workloads, and how to do that efficiently. So I think this is valuable internally to Microsoft, but we could also see the value to the broader community, and I think that’s why we decided that this should be done completely in the open.
Now, as far as “Did things like this exist?”, not to my knowledge. A few months ago Brendan Burns and a couple of other people put together a prototype of something like this to connect ACI to Kubernetes, kind of proved out the concept, and we kind of decided to take that and turn it into a much more fledged-out product, with more features, and a community effort.
I think there’s some stuff for doing serverless with Kubernetes; correct me if I’m wrong, Brian, but I can’t remember the name of the project… There’s one out there. But I think we saw this as kind of more of — so serverless I think was containers; you have the warm-up time of the container and stuff, and I don’t know whether we’re quite there yet, but definitely the batch and CI/CD jobs and bursting out into a cloud provider – I think that that’s the main appeal in the core use cases we’re focusing on first.