I'm getting closer to deploying my Pinot to a production stack. I want to run a load test to better understand how to deploy and configure Pinot. Any advice for the load tests? I want to load 50B rows of sample data and (1) measure latencies and (2) iterate on the schema to optimize latencies. I figured I'd deploy a one off AWS cluster for this and iterate on latency tests against it. How many servers do you think I'd need for this? 50B rows. Probably about 10-20 number fields per row.

start with as many nodes required to store the data