Size an API gateway's request ceiling from per-request CPU cost, with uplifts for authentication, transformation and cache hits.
An API gateway spends a near-constant amount of CPU per request on parsing, routing and policy evaluation, so its ceiling is the available core-microseconds divided by that cost. Cache hits skip the upstream call and most of the policy chain, so they are blended in at a fraction of the full cost. JWT validation and body transformation together can double the per-request cost, which is why a gateway benchmarked with a pass-through route delivers half that number once real policies are attached.
ADC API Gateway Throughput
effective CPU = base CPU × auth uplift × transform uplift, blended so cache hits cost 15% of a full request; max rps = cores × 1,000,000 µs ÷ effective CPU.
effective CPU = base CPU × auth uplift × transform uplift, blended so cache hits cost 15% of a full request; max rps = cores × 1,000,000 µs ÷ effective CPU. An API gateway spends a near-constant amount of CPU per request on parsing, routing and policy evaluation, so its ceiling is the available core-microseconds divided by that cost. Cache hits skip the upstream call and most of the policy chain, so they are blended in at a fraction of the full cost.
JWT validation and body transformation together can double the per-request cost, which is why a gateway benchmarked with a pass-through route delivers half that number once real policies are attached.
This calculator takes 7 inputs: Requests per second, Average response payload, Base CPU per request, CPU cores available, Validating a JWT or API key per request, Request/response transformation enabled, Responses served from the gateway cache. The pre-filled defaults are a realistic starting point — replace them with figures from your own environment for a result you can act on.