The scripts in this repository enable users to create base images for their models, as well as model container images built using these base images. The containerized model images can be validated and deployed as containers using Kubernetes with these tools. The Docker Container Registry and Google Container Registry cloud platforms are supported, and we are working toward supporting other cloud platforms, such as Amazon Web Services and Azure.
Note: For information about creating base images for use with either SAS Model Manager or SAS Open Model Manager, see Create Base Images in the Open Model Manager Resources GitHub repository.
Here are the products and solutions that are currently supported:
- SAS Viya
- Single installed Linux machine
- Single tenant
- Python 3.4+
- Cloud Provider -- Google Cloud Platform, Amazon Web Services, Microsoft Azure
- Base images for R and Python 3
It is recommended that the scripts be downloaded onto the same machine where SAS Viya is installed, and that the system user be set to sas instead of root. You can set the user to sas using the following command:
$ su - sas
The system user must have Python 3.x installed and set in their environment path. You can check the version of Python that is running with the following command:
python --version
In addition, use pip to install the following packages for Python 3:
$ pip install docker --user
$ pip install kubernetes
$ pip install python-slugify
You can check if a package has already been installed by executing the following command:
$ pip show <package>
The system user must also have Docker Community Edition. To install, use the following commands:
$ sudo yum install -y yum-utils \
device-mapper-persistent-data \
lvm2
$ sudo yum-config-manager \
--add-repo \
https://download.docker.com/linux/centos/docker-ce.repo
$sudo yum install -y docker-ce docker-ce-cli containerd.io
$sudo systemctl enable docker.service
$ sudo systemctl start docker.service
If a 'docker' group does not already exist, create one using the following command: Note: The previous install commands should have created a 'docker' group.
$ sudo groupadd docker
Add the 'sas' user to the 'docker' group
$ sudo usermod -aG docker sas
Download the source code from the Model Container Recipes Git repository.
If the SAS Viya home directory is not /opt/sas/viya, edit the property value viya.installation.dir in the config.properties file. If it has not already been done, switch the system user to sas with the following command:
su - sas
We currently support 3 different Cloud Providers for use with our model container recipes. For information about configuring your Cloud Provider and Cloud Settings properly, please refer to the appropriate configuration guide from the options below:
If this is your first time using the utility, you must complete the required fields within the config.properties file. This file is located within the following directories: base-image-cli and model-image-cli. For more information, see the base image folder and the model image folder.
If you have already configured your properties files, learn how to create a base image or create and use a model image by reviewing the respective READMEs.
This project is licensed under the Apache 2.0 License.