Support GPU instances on APPUiO Cloud Exoscale

[maetthu-indermuehle]

We currently have dedicated Nodes at Exoscale that have GPUs. Some of our customers are using that currently. According to the discussion with our engineers on the current dedicated nodes setup we just added nodes with GPU without any other changes to our deployment and it worked. Of course without dedicated nodes this is not that easy anymore and we would have to tag these nodes in a way and have the customer tag pods to run on these nodes. Furthermore the billing would have to take that into consideration (for example with a higher per GB Ram Price).
It would be interesting to see if that is a product we could offer. From my point of view I see two advantages:

• Offer the current customer a solution.
• Have a unique feature for our platform that no one else has (from my quick research) on a public platform.

I also found an interesting blog entry on RedHat about a related topic: https://cloud.redhat.com/blog/multi-instance-gpu-support-with-the-gpu-operator-v1.7.0