How gatling measures latency of request?

Given the fact that our client working with connection-close, so each request consists with the following sequence:

  1. → Syn
  2. ← Syn-ack
  3. → Ack
  4. → Push (http request)
  5. ← Ack
  6. ← Push (http response)
  7. → Ack
  8. → Fin
  9. ← Fin-ack
    What is the request duration measured by Gatling? Between what packets exactly?


Assuming no reuse of existing keep-alive connection, that’s basically it, except it might also include hostname (DNS) resolution if host wasn’t already in the cache.