Hi!
I'm setting up 2 servers with Passenger, serving a single web application.
The documentation suggests at least the same amount of passenger instances as available cores, to increase concurrency. I agree that this is the bare minimum you can do to get good concurrency.
Memory consumption is not really an issue for us. Our servers are trivial to scale up memory-wise.
So I'm wondering, if memory is not an issue, is there a higher baseline I can use for my max pool size? Would cores x 2 still increase concurrency (even if not linearly), or am I already in the zone where I'll be context-switching too much?
Thanks a lot,
Mat