Run:ai SaaS Cluster installation uses the Run:ai cloud as its control plane. The cluster sends information to the cloud for the purpose of control as well as dashboards. The document below is a run-down of the data that is being sent to the Run:ai cloud.
If the data detailed below is not in line with your organization's policy, you can choose to install the Run:ai self-hosted version. The self-hosted installation includes the Run:ai control-plane (also called "backend") and will not communicate with the cloud. The self-hosted installation has different pricing.
Following is a list of platform data items that are sent to the Run:ai cloud.
|Job Metrics||Job names, CPU, GPU, and Memory metrics, parameters sent using the |
|Node Metrics||Node names and IPs, CPU, GPU, and Memory metrics|
|Cluster Metrics||Cluster names, CPU, GPU, and Memory metrics|
|Projects & Departments||Names, quota information|
Run:ai does not send deep-learning artifacts to the cloud. As such any Code, images, container logs, training data, models, checkpoints and the like, stay behind corporate firewalls.