Hi folks!
I want to host Prefect on GCP for production and have a question regarding the architecture. I'm thinking of three solutions.
• Hosting it on a VM -> Start VM whenever needed and run the flow, and then stop it.
• using Cloud Run and hosting Prefect in a serverless way.
This is just for the UI and metadata and logging, right?
To run the flow and deep learning stuff, we need to use another machine with GPU or GKE or ...?
What do you think are the pros and cons of these solutions? Any other recommendations? Any proposed architecture?