Metal DB - Using Hardware like a POD

What if you could schedule and use bare metal hardware as easy as a pod in Kubernetes or OpenShift? Could you really make significant performance gains in applications, that are totally driven and optimised automatically?

The components to do this, exist today. Using Metal3, which takes hardware like the above, and makes it act just like a Kubernetes object like a pod, we can dynamically schedule physical hardware and move apps up and down multiple types of hardware totally dynamically. How can we use such a thing? Using the MetalDB Operator driving Metal3 Operator, which then can manage physical boxes.

Image a Kubernetes Operator that understood the ebb and flow of your business, that knew where your physical compute assets were, and could make sure there are replicated copies distributed for redundancy on a "Cold" tier to make sure you always have it, could then move it up to Warm tier as it started getting used, and as the season comes around could move it up further to more performant hardware as the peak hits. As things calm down it could scale it down and free up resources for more important databases.

MetalDB is a design Im working on, that combines Database knowledge and Hardware architecture together. Its a native Metal3 application which using OpenShift to manage a large scale bare metal database machine designed to merge data warehouse, mid-tier database, and high performance database operations all-in-one.

The MetalDB Operator can use either NuoDB or CockroachDB as the underlying SQL engine. The intention is for MetalDB to give a Database Administrator the ability to better manage and implement complex multi-machine rack-scale accelerated applications with far better utilisation and cost savings. Its design to reduce or remove lock-in by major database companies, and let your applications to scale and be more secure.