[mack1001]

Days are numbered for infra software detached from a cloud provider. Cloud Foundry and VMware folding (partnering with Google Cloud) is a good indicator that there is no clear path forward. So Kubernetes on prem or custom rolled on a cloud will not win the fight against GKE or ECS. Openstack,DC/OS, Openshift are all risky bets from a career standpoint.

[rdtsc]

Or is it the other way, days of overcharging and being locked into a cloud provider are numbered since the software that runs the infrastructure is not the secret / proprietary anymore.

See, can spin it either way :-)

[mack1001]

Small data point but the uptake of aws lambda has been crazy in spite of the clear lock-in. Similarly S3 and several other proprietary cloud services. The cost of infra software is nominal compared to what it takes to manage it and maintain high uptime.

[moondev]

The uptake of k8s has been way more prominent than lambda, and k8s is platform agnostic. That's the whole reason AWS dosen't offer it as a managed service.

[anoother]

Some people want to know how the nuts & bolts work as opposed to putting together borrowed Lego blocks. Linux has more than proven the utility of diversely-sourced development, and Kubernetes benefits from the same right now. I'd be worried if that changed.

Also, there are (and will continute to be) plenty of companies/projects that can't use public clouds for a variety of security, compliance, performance and business inertia concerns ('concerns', not necessarily 'reasons').

[mack1001]

In the next few years, on prem will become the exception than the rule considering that heavily regulated industries are now moving to the cloud. The same way hardware providers like HP, DELL and EMC are struggling, IMHO the next wave will be infra software providers. Unless for very specific purposes, the ROI of running your own infra software is decreasing.

Edit: typo