A thread-based http server is limited by the number of threads it supports (either with a static maximum or with the possibility of dynamic increments of thread count), particularly when each request is synchronously long-running (e.g., long delay times in accessing databases and remote/cloud storages).

So if I migrate the same code over to nginx, how does it handle the long-running nature of the code in an nginx module's handler, if it is to handle ~10k requests simultaneously using a single (or a few) thread(s) as nginx's architecture claims?