JAjacques_chester·Jun 11, 2017
A first note, this is not like-for-like. Dokku, Flynn and Deis fit the PaaS spot more meaningfully. Kubernetes and Swarm are sometimes called "Container as a Service" -- you still need a bunch of work to select, install and configure features that PaaS users take for granted.
Missing from the list: Cloud Foundry. And OpenShift! Give Red Hat some respect here for making Kubernetes into a platform. And DCOS, which I think got its thunder stolen by Kubernetes.
But I mostly know Cloud Foundry, because it's my day job to keep making it better.
It's the most mature out of all of these, has been soak tested to 250,000 running applications[0], can be deployed to any major IaaS or bare metal, comes with routing, logging, service injection, healing, no-downtime upgrading and I forget what other headlines I usually pick out of the several hundred fe
Read the rest
atures it now includes.
The Cloud Foundry Foundation includes Pivotal (my employer), IBM, SAP, Google, DellEMC, VMWare, Cisco, Suse and those are just the fancy tech names.
The reason you don't hear about it is that we and our partners have focused on competing for enterprise sales. It's not how you get publicity on HN, but it does mean that PivotalCF -- our commercial distribution -- has the fastest-growing sales of an opensource-based product in history. And sales are still zooming up and to the right.
The nice thing is that we and other partners get very broad, specific feedback from customers who are already at massive scale and who expect utter, non-negotiable reliability. We run a public service ourselves (Pivotal Web Services), SAP have HANA Cloud, IBM runs BlueMix.
We and other Cloud Foundry contributors have had the benefit of that dogfooding and feedback for longer than any other container-based platform in existence. And it turns out, there are so many things that can go wrong. So many. It's crazy.
Lastly, Cloud Foundry teams have massive investments in project automation. This gives us two capabilities: one is that we can roll out complete rebuilds of the whole platform within hours of an upstream CVE patch. Users can then apply these fixes to their platform live, in-place, without any running apps noticing that it has occurred. BOSH[1] will roll out the deployment with canaries, and Diego[2] will relaunch containers as they are reaped during upgrade.
The second capability is that we are confident in making very deep, very aggressive changes if it proves necessary, because we have tests upon tests upon mountains of more tests. And again: nobody notices, except that their platform gets faster or becomes reliable under even more extreme circumstances.
If I sound like an utterly biased fan, it's because I am an utterly biased fan. I've watched this thing evolve up-close for years. It is amazing.
We build installable superpowers.
Disclosure: If it's not obvious, I work for Pivotal on Cloud Foundry.
[0] https://content.pivotal.io/blog/250k-containers-in-productio...
[1] http://bosh.io/
[2] https://github.com/cloudfoundry/diego-design-notes
Minor edits for style.