Red Hat Expands Scope of AI Ambitions for OpenShift Platform
Red Hat this week expanded the scope of its efforts to enable organizations to build and deploy artificial intelligence (AI) workload on an instance of its OpenShift platform based on Kubernetes.
In addition, Red Hat is simultaneously making available a technology preview of an instance of the Red Hat Enterprise Linux (RHEL) operating system that can be deployed as a container image to streamline DevOps workflows.
Red Hat Summit 2024
Announced at the Red Hat Summit 2024 conference, Red Hat is now making it possible to use KServe, a Kubernetes custom resource definition that enables Kubernetes clusters to orchestrate multiple types of AI models. In addition, Red Hat is adding support for a library for deploying inference engines for large language models (LLMs), dubbed vLLM, and a Caikit-nlp-tgis runtime for natural language processing (NLP) models and tasks.
Red Hat is also adding support for the open source Ray framework for running Python workloads across multiple clusters and CodeFlare, an open source serverless computing framework that optimizes consumption of IT infrastructure for AI models.
At the same time, Red Hat is now, in technology preview, making it possible to deploy an AI model on a single node cluster typically used for applications deployed at the network edge.
Red Hat is also tightening integration with development tools such as VS Code, RStudio, via a technology preview, and the CUDA framework and a NIM microservices framework for inference engines developed NVIDIA. IT teams can also now configure different types of hardware accelerators and take advantage of existing Red Hat tools to monitor AI models.
An extension for Podman Desktop, dubbed Podman AI Lab, also now enables developers to build, test and run containerized generative AI applications on their local workstation. Designed to work with the container image mode of RHEL 9.4, it also includes recipes for building applications such as chatbots that developers can extend.
Red Hat Extends Alliances
In addition, Red Hat is extending its alliances with Intel and AMD to make it simpler to deploy Red Hat OpenShift across multiple classes of processors, including graphical processor units (GPUs), and accelerators. It will also become simpler to optimize AI workloads running on instances of Red Hat OpenShift running on GPUs via an alliance with Run:ai, a provider of an orchestration tool recently acquired by NVIDIA. Red Hat is also partnering with Stability AI to make it simpler to deploy open AI models on Red Hat OpenShift platforms.
Steve Huels, vice president and general manager for the AI Business Unit at Red Hat, said the overall goal is to make it simpler for IT teams to build and deploy AI models across a hybrid cloud computing environment as they best see fit.
As part of that effort, Red Hat later this summer is committing to adding support for large language models (LLMs) to Konveyor, an open source toolkit for migrating source code to a Kubernetes environment.
Red Hat also extended its alliances with Oracle and Pure Storage to make it simpler to deploy Red Hat OpenShift in the cloud and on-premises IT environments.
Additionally, Red Hat is making a developer preview of Red Hat Connectivity Link available. Based on open source Kuadrant software, Red Hat Connectivity Link provides a framework for application connectivity spanning multiple cloud computing environments that unifies the management of ingress gateways, service meshes, load balancers and application programming interfaces (APIs).
Red Hat late this year will extend the reach of its existing Lightspeed generative AI platform for automating IT operations to the Red Hat OpenShift platform and plans to make a similar capability available for RHEL at some unspecified date.
Finally, Red Hat is making available an optional additional 12-month EUS term for OpenShift 4.14 and subsequent even-numbered Red Hat OpenShift releases in the 4.x series to take the full lifecycle available for these releases of OpenShift to three years.
It’s not exactly clear how many organizations have standardized on Red Hat OpenShift to build and deploy cloud-native applications. The one clear thing is that as Kubernetes increasingly becomes a de facto standard for building and deploying AI applications, Red Hat sees a major opportunity to advance its approach to hybrid cloud computing.
