Why is docker not building a Serverless CaaS offering?

[torquelab]

Running containers across cloud providers is a pain at this moment. Different providers have their own ways of running the containers through their CaaS offerings. Many providers don’t support scaling to zero. Pricing is based on multiple parameters and most of the time same compute on a CaaS offering cost 3-4 times of the VM with similar specs.

Many (small) startups are not interested in kubernetes because of its complexity and number of people required to manage it. So, the need of the hour for such startups is a simple CaaS offering with simple pricing and ability to scale to zero at a minimum. Cloudflare recently released the ability to run the containers in the production environment. However, it seems to me that docker has a good opportunity here to make an impact considering the fact that most the developers are using docker in their development environments for containerized applications.

Creators of MongoDB has made it very simple to run it across a number of hyperscalers through the Atlas offering and IMO docker can build a similar offering making it possible to run containers in a unified manner across a number of cloud providers. This is not possible at this moment unless one is using self-managed k8s which as I mentioned earlier is not the first priority for small startups.