For improved stability, scalability and maintainability, we want to spike the deployment of the Prometheus Agent as a daemonset with one instance on each node of the spoke.
Each instance is in charge of scraping itself the metrics from the targets located on its own node.
This enables:

• Scaling naturally with the number of endpoints, no need to tune resources requirements.
• Independence from the spoke monitoring solution

This requires mounting the needed secrets and permissions for accessing platform metrics endpoints to scrape, access the service monitors.