Seleziona la tua lingua
Red Hat is working on the next release of the supported enterprise distribution of OpenStack, Red Hat OpenStack Platform 15, based on the Stein community release. In this multi-part blog series, we’ll be examining some of the features that Red Hat and the open source community have collaborated on–starting with a look to future workloads, such as artificial intelligence.
"How does OpenStack enable next generation workloads?" you ask. When it comes to computer-driven decision making, machine learning algorithms can provide adaptable services that can get better over time. Some of these workloads, such as facial recognition, require GPUs to ingest and process graphical data in real time. But the more powerful GPUs often used for machine learning and such are expensive, power-hungry, and can take up a lot of room in the servers' chassis. When working with GPUs at scale, optimized utilization is key to more cost effective machine learning.
To help make the most of the money spent on GPUs, Red Hat has virtual GPU (vGPU) support in its roadmap for OpenStack Platform 15. vGPU support allows GPUs to be virtualized, which allows multiple applications or users to leverage the same shared, scalable, managed resources in a software-defined world. The feature is in tech preview in Red Hat OpenStack Platform 14 already.
When Red Hat OpenStack Platform 15 is generally available, the vGPU feature can be used to run workloads, like the TensorFlow open source machine learning framework, that will comprise multiple virtual machines (VMs) leveraging a single GPU. Imagine a single GPU providing instant, local face recognition for airport security gate VMs or always-ready quality checks for just in time (JIT) manufacturing facilities. Fast, intelligent, highly available services that can scale!
Want to see how it all works for yourself? Check out Red Hat Product Manager Erwan Gallen’s own blog to see the nitty-gritty. It’ll walk you through the details of enabling vGPU across your clusters so you can see for yourself how Red Hat and the community innovate together.
At Red Hat, we work closely with the open source community to provide the open toolset for IT and we love to see the groundbreaking, cloud-powered applications that our customers run. Stay tuned for our next post about another interesting feature from Stein.