Enhancing OpenShift Data Science support for on-premise deployments
We’re delighted to announce the GA of a self-managed software version, following the general availability of the Red Hat OpenShift Data Science cloud service a little more than a month ago.
An easy-to-configure MLOps platform for creating and deploying AI/ML models and intelligent applications is Red Hat OpenShift Data Science (RHODS). Data scientists may train models utilizing notebook pictures and frameworks like PyTorch and TensorFlow, among others, with the core environment provided by OpenShift Data Science, which has access to both software-defined CPU accelerators and GPUs. Additionally included in this updated version of OpenShift Data Science are the improved model serving capabilities released in December. Customers can implement other open-source AI/ML technologies as well as their own notebook images and optional partner technologies in the RHODS environment.
Pre-integration of many optional technology partner offerings is offered by Red Hat OpenShift Data Science, including Anaconda Enterprise, IBM Watson Studio, Intel OpenVINO, AI Analytics Toolkit, Pachyderm, and Starburst Galaxy (for the cloud service). Additionally, RHODS can be utilized in combination with any of the more than 30 AI/ML software partners who have certified OpenShift operators.
While the cloud service offers a more hands-off IT operations experience, especially for customers who prefer to entrust Red Hat SRE professionals working with our cloud partners like AWS with the application platform infrastructure and tooling updates, a self-managed version appeals to those who:
- Want to keep the entire process of preparing data, developing models, and deploying those models near to the data, on-prem or even at the enterprise edge
- Are prohibited from cloud deployment by compliance requirements
- Are OpenShift users, Red Hat’s top application platform, but have not yet adopted Red Hat cloud services (OpenShift Dedicated or Red Hat OpenShift Service on AWS)
Our strategy is to maintain the same functionality throughout the self-managed product and cloud service. To maintain the self-managed version’s release cycles in step with the cloud service, we also intend to iterate the version quickly. There will undoubtedly be a few minor differences. For example, in the self-managed version, air-gapped support for disconnected environments will be made available in late January. This feature will be appealing to government clients and other users with connectivity requirements.
The fact that Red Hat offers both a managed cloud service and a traditional software AI platform, in our opinion, is consistent with the company’s focus is giving customers a choice. Models can be created in the cloud and deployed on prem. Create models prem and release them to the cloud or edge. Your decision. The flexibility of OpenShift (and now OpenShift Data Science) to run on various footprints offers a hybrid MLOps environment that jointly connects IT, data science, and application development teams together in addition to offering a powerful application platform powered by Kubernetes.
Customers are relieved of the responsibility of independently connecting these AI/ML technologies and maintaining their various life cycles thanks to RHODS. It makes it easier for IT Operations to manage the platform by letting platform engineers design configurations for their application developers and data scientists that can be scaled up or down and managed with minimal effort.
The ease with which IT Ops can manage the environment for their data science and application development teams has been praised by early customers. They have said that makes it easier to tailor the solution to meet their needs to the admin UI capabilities. Early users loved being able to easily set up their users using familiar data science tools and their own personalized notebook pictures.
Don’t rely on what we say. Take a test drive. For data practitioners who want to get their feet wet, we have a developer sandbox equipped with sample learning routes.