I know some will say, this depends on your test scenario & environment and that you need to run your own tests to determine. But it would be nice to get some industry average ball park figures.

What I'd like to ask is typically for running a decent load against production environments, like say a e-commerce website, how large of a JMeter farm/cluster/node config did you need to have?

I'm assuming in 90-99% of the cases, one or a small few JMeter nodes was not adequate for the load generation (between 10-300 threads per node, but typically 50-100 or less threads). Unless being a novice, I'm configuring & using JMeter (mainly on Windows) incorrectly.

Just wanted to know because a large node set would require quite a test lab environment or use of cloud virtualization to have the required hardware for the load generation.