3 ms·
Thread per request works because the GIL is released during blocking IO. You only need one process per core and each process runs multiple threads. It saves a m
by jashmatthews 8y ago
Thread per request works because the GIL is released during blocking IO. You only need one process per core and each process runs multiple threads. It saves a massive amount of RAM compared to trying to cram 50 (g)unicorns onto a server.