Just to be clear (not sure if you're kidding) but people use Passenger (and production application servers in general) not just because they meet some minimum throughput threshold.

Passenger takes care of keeping your application alive, makes sure slow clients don't break it, automatically handles concurrency and has low latency. That's just the basics, there's a lot of configuration options for the more serious deployments.

Sure, but at a few thousand visitors a month that clocks in around one request every 10 minutes or so. I'm not saying they shouldn't use Passenger or another production app server (I use Unicorn fronted by nginx personally), but it'd honestly probably be fine with Webrick. I'm also not saying that Webrick is what they'd want to be on since it doesn't really have any supervision or real smarts.

With that few requests, the wrong app server will terminate the process, meaning that every request will endure massive startup costs.

It also probably means that the server is on a $5 VM, so memory pressure is also a concern.

How does Passenger compare to unicorn? (I'm curious)

Passenger and Unicorn (from Passenger): https://github.com/phusion/passenger/wiki/Unicorn-vs-Phusion...

Passenger and Puma (from Passenger): https://github.com/phusion/passenger/wiki/Puma-vs-Phusion-Pa...