Infrastructure Provisioning

We use docker a lot - in production as well as in development environments.

With that said, the current decision is to deploy production stacks with ansible, on bare metal.

The rationale behind this decision is that, in disaster recovery, it may be as difficult to setup the docker engine, as to just install the app on metal. Besides, an application will have more than one foreground process, requiring more than one container.

There is nothing wrong with deploying production stacks on docker. At Wasya Co we do that all day long, for example, all Drupal installations are dockerized. However, for more complicated applications, such as ruby-on-rails stacks, and specifically for disaster recovery from scratch, installing on metal with ansible may be no more difficult, and no less stable, than installing docker and then multiple containers.

Of mention, Amazon provides Fargate capability to deploy docker containers. We have deployed stacks on that infrastructure, however, doing it this way is slightly more complicated than deploying your own containers or your own bare-metal stack. As well, using Fargate locks you into the amazon ecosystem, whereas if you install docker engine from scratch, you are free to chose your cloud provider (such as Lumen).

Please log in to post comments:  
Login with Google