09-18-2018, 10:43 AM
I've got a Clusterboard and 7 SOPine modules sitting in my cart with the excitement of thinking I will run Mesosphere DC/OS and/or Kubernetes on it to build out, literally on my desk, a fully elastic containerized microservices orchestrated backend (FECMOB) for my mobile/web app. And then in theory once I've written it all down from a composer and orchestrator perspective I could seamlessly play the whole systematized symphony out on the global cloud providers. (I think of it as going from "hyperlocal fog" to "global cloud") 
But then I have a huge question; is the performance of this hardware gonna make it unusable for doing such a thing? Is a single-board-computer-cluster (SBCC) just a geek toy or is it performant enough for a developer to actually use in building a real cloud application? I know any containers I run on this thing will be be slower than the running them on the "real" thing (actually they're all containers though too aren't they) but the question is how slow? Does anybody have a sense if this would work or if the whole system will just be so slow and frustrating given the inherent performance of the Cortex A53 with 2GB of RAM that I shouldn't even attempt this to avoid the disappointment? On spec it looks like this board gives you hardware that roughly equivalent to 28 t2.nano instances but they're ARM vs XEON.Thanks
