Event Loop Delay Threshold Affecting Request Latency in Vert.x

Determine the point at which the Vert.x event loop delay metric begins to translate into measurable increases in end‑to‑end request latency when handling concurrent HTTP requests.

The observable latency depends on factors such as JVM safepoints, OS scheduler granularity, and the configured worker pool size, none of which expose a fixed limit for the delay threshold. Understanding this relationship would help size worker pools and tune blocking‑offload strategies.

What event loop delay value marks the onset of latency degradation? How does the threshold shift when the worker pool size is increased or decreased? Can the threshold be estimated from JVM safepoint frequency or OS tick rate under a given load?