What are the maximum throughput of VRT services single instance?
It depends based on the configuration of VRT.
However, the following can be done to measure the performance.
Create a VRT instance with single replicas and desired configuration, I suggest use 2GB for IGW and ESB and 1 GB for others and based on busines use case measure the throughput.
If we consider our sandbox/CUG environment, how many service requests does it process per second?
In the range of 8 to 10 Requests / Second with a payload size of 50 KB