What is the recommended Xmx value for running the API service?
The 2GB setting on the On-Premise "quickstart" distribution is a default for development usage. In production, we recommend putting as much memory as you can.
Essentially, the API should use as much memory as is allocated to the API docker container. We usually ask customers to consider an M class large instance type (something on the order of 7.5GB of RAM per API).
The UI can use considerably less. For the UI, a 2GB allocation is fine for production.