Nvidia pulls sheets off EGX, an edgy machine learning platform based on Kubernetes

All kinds of software is moving into lightweight application containers, and the fact hasn't been lost on Nvidia – the GPU supremo has launched a platform for machine learning at the edge of the network, based on Kubernetes. EGX is compatible with all mainstream devices made by Nvidia, from the tiny Jetson Nano boards used for …

[Alan Sharkey]

Confused

Is it just me? I thought Kubernetes was an orchestration management product. What's that got to do with machine learning?

Alan

[Peter Galbavy]

Re: Confused

Hipsters

[Bronek Kozicki]

Re: Confused

A lot. Your ML workloads will likely run on K8s, if you want to reuse the infrastructure bits (such as deployment, resources scheduling, logging, monitoring etc.) you have for all other K8s workloads already. The trouble with K8s and ML is that GPUs are another type of resource which needs to be supported by orchestration (see resource scheduling, above) and containers need support for the GPU driver access.

[JohnMartin]

Re: Confused

K8s provides a nice set of infrastructure controls that helps you to layer on other things that accelerate the entire machine learning lifecycle (https://en.wikipedia.org/wiki/MLOps ).

One example might be run.ai, (https://www.run.ai) or in the case of my employer, the Data Science Toolkit (https://blog.netapp.com/streamlining-ai-workflows-with-data-science-toolkit) which builds on top of that.