Frequency of API requests bucketed by response latency (10ms to 500ms), peaking around 50ms with a long right tail. A red `lineY` annotation marks the request volume sitting at the 200ms SLA budget so slow-request mass past the cutoff stands out.