Estimate reverse-proxy throughput, worker CPU utilisation and maximum request rate, with the TLS termination penalty included.
A reverse proxy costs a roughly fixed amount of CPU per request plus per-byte copying, so dividing the available worker seconds by the per-request cost gives the request ceiling. TLS termination adds around a third to that cost even with hardware AES. Keepalive ratio is the hidden variable: dropping from 90% to 50% reuse triples the new-connection rate and pushes the handshake path, not the proxy CPU, into being the bottleneck.
Reverse Proxy Throughput
throughput = requests/s × response size × 8 ÷ 1000 Mbps; worker utilisation = requests/s × CPU per request ÷ workers; max rps = workers ÷ CPU per request.
throughput = requests/s × response size × 8 ÷ 1000 Mbps; worker utilisation = requests/s × CPU per request ÷ workers; max rps = workers ÷ CPU per request. A reverse proxy costs a roughly fixed amount of CPU per request plus per-byte copying, so dividing the available worker seconds by the per-request cost gives the request ceiling. TLS termination adds around a third to that cost even with hardware AES.
Keepalive ratio is the hidden variable: dropping from 90% to 50% reuse triples the new-connection rate and pushes the handshake path, not the proxy CPU, into being the bottleneck.
This calculator takes 6 inputs: Requests per second, Average response size, Worker processes or cores, CPU time per request, Terminating TLS at the proxy, Requests reusing an existing connection. The pre-filled defaults are a realistic starting point — replace them with figures from your own environment for a result you can act on.