On performances and profiling #654
Description
This discussion does not regard features or improvements to the deployment, the SDK, ... It is around scalability and infrastructure requirements.
We have observed that in many occasions Chall-Manager consumed a lot of resources when deploying instances.
This will be confirmed experimentally through the 24h IUT 2025 monitoring.
It would be interesting to monitor profiling info in live such that resources consumption could be tracked down -a bit. It might not be sufficient as Pulumi runs extra processes, but it could the way to understand whether performances issues comes from our work or is tied to something else.
One approach would be to add a signal exporter with OpenTelemetry.