In the testing guides, it mentions that “A single instance of k6 is often enough to generate load of 30.000-40.000 simultaneous users (VUs).” Is there a reason as to why it is specifically 30k to 40k? Other than hardware limitations, in theory, wouldn’t k6 be able to generate an unlimited amount of VUs?
Well, in practice, hardware limitations exists… 
Even for 30k-40k VUs on an instance, you’d have to have a relatively expensive machine. At some point it becomes very impractical or impossible to scale your hardware vertically, since a bunch of smaller machines will be much cheaper, especially using cloud providers.
I think you should split the VUs into the small machine , because if you simulation 30-40k VUs in 1 , will take a long time to simulate this and cost a lot of money , the value of K6 that you can simulate this and take a little resources for server . So don’t try to cost to much
1 Like