I use Lumen to receive CallBacks from a service which I need to answer 200 within 200ms. Performing tests manually I noticed that when I send a callback sequence the first one always takes much longer to return, the rest return super fast. This is generating a problem regarding the reception of calbacks from the service I use.
Is there something particularly different about that first callback?
It is difficult to diagnose anything blindly; it could be literally anything, e.g. cold caches.
I would try to isolate if the latency is coming from the Lumen application in the first case; then narrow it down to specific units of functionality if necessary.