Docker Images / Containers
Each ScienceOps model runs inside of its own docker container. That container is based upon a custom image that is built for that model.
If you're not familiar with docker, you can think of an image as a mini operating system with a custom set of components (files, packages, configs) that are required to execute a model. For example, if an machine vision model is deployed that requires OpenCV, then the model's image would contain the C/C++ bindings for OpenCV. That image is then run as a container, which you can think of as a running application (in our case a Python or R model).
You can customize the docker images used by ScienceOps to include commonly used packages, proprietary packages, database drivers, or other use-case specific components needed to execute your company's models. There are 2 ways to customize the docker images your model(s) will use: