Everybody gets a Kubernetes-as-a-Service!
You can’t go anywhere without someone is offering you a Kubernetes-as-a-Service. It's convenient, a one-click cluster, don’t worry about masters, grow and contract on demand, consume VMs out of the standard catalog to run your K8s apps. Phew! I’d never build another Kubernetes cluster in my life. Until, the idea of building a KaaS by yourself becomes real!
Fans of datamattsson.io knows I’ve been touting the Cluster API and true multicloud as enablers and drivers for KaaS. K8s clusters is the new cattle in the industry, vend, dispense and dispose at will. New app? New test, stage and prod cluster — done and dusted.
Anything-as-a-Service
I recently enjoyed reading the CEO’s of Packet blog why there is no such thing as “Packenetes” delivering a MKS (Metal-K8s-Service). Packet business model differs a lot from the big three so I’m not sure how relevant the comparison is. Anyhow, the glimmer of insight I caught from reading the blog is that software will outsmart Anything-as-a-Service (a lot of a's) if it becomes popular. Again, looking at the Cluster API proves that building better APIs and consumption models for infrastructure is the way to go.
More-as-a-Service
While browsing for something completely unrelated I found another project that have flourished in this space, good old Docker Machine and the ever growing list of machine drivers. Dispense a Docker Engine at will anywhere, cobble together with Swarm or bootstrap a kubelet and voilà, BYO KaaS!
Everything-as-a-Service
With HPE Nimble Storage dHCI you can standup a KaaS capable platform in less than an hour, expandable and contractable at will in any direction and deploy another one the next day. First KaaS platform out is of course Google Anthos GKE On-Prem. But, as the dHCI platform expose the vSphere APIs, it’s also ready to dispense Docker Machines among a few other things. :-)
It goes without saying that HPE made the bold statement that everything (Everything-as-a-Service) we do will be offered through a consumption based model through HPE Greenlake by 2022.
Now, back to work.
