Containerization is a key technology that accelerates the adoption and impact of data science amongst numerous companies in the world. An experiment’s state (data, code, results, package versions, parameters, etc.) at any point in time can be captured as a whole in a lightweight container and deployed on customers’ on-premise infrastructure and/or elastic compute infrastructures such as the cloud. This means data science experiments are easy and fast to reproduce, and data science teams can build on the prior work of others, which ultimately helps them drive the next level of innovation.
Now with SAS Analytics for Containers on Domino, SAS users can reap all the benefits of containerization and the cloud with SAS analytics and SAS Viya. These pre-built SAS containers can easily be imported into Domino’s elastic compute infrastructure for data scientists to use, allowing them to cross-pollinate with their other experiments in R, Python, or using TensorFlow, H2O, etc. with SAS. Using SAS Analytics for Containers on Domino, users can write SAS code, develop models, and publish applications by directly using SAS Studio in the browser or in a Jupyter Notebook. Users will no longer need to use a R or Python wrapper for SAS. Instead they can benefit from the native SAS experience, while taking advantage of all the experiment management capabilities provided by Domino including reproducibility, collaboration, and easy access to scalable compute. Furthermore, Domino allows IT to manage SAS and open source workloads all in one place and helps IT teams navigate the migration of SAS to the cloud.
SAS Analytics for Containers on Domino provides a clear path for organizations to accelerate their adoption of cloud platforms like AWS allowing organizations to reduce the size of their data centers, take advantage of elastic compute for inherently bursty workloads, and moving all analytic workloads to a centralized infrastructure. Using Domino Lab, data scientists can readily run SAS Analytics and Machine Learning workloads with their preferred compute resources in isolation with a single click, without having to wait for other workloads to finish. Additionally, data science teams can spin up parallel experimentations for faster iterations, modeling tournaments, and results.
Domino tracks all versions of SAS projects and captures each experiment’s state, including data, code, SAS version, environment, discussions, parameters, and results. These projects can be shared amongst data science teams for collaboration and as the foundation for future projects. Capturing these artifacts helps users achieve model provenance and governance and provides auditability in their workflows.
When data scientists are ready to deploy their SAS models, they can take advantage of the Domino Launchpad capabilities to easily publish models (and different versions of the models) on a scalable compute infrastructure. Hundreds of models can be deployed and run simultaneously.
Domino uses containers as the foundation for its model management platform, which can be deployed on-prem or in the cloud. Furthermore, Domino has enabled users to use any tools of choice, whether open source or proprietary. With the addition of SAS Analytics for Containers on Domino, users have SAS data science tools at their fingertips without having to worry about documentation or reproducibility. All aspects of model management are taken care of by Domino, allowing users to focus on what they do best: creating SAS models and driving innovation.