Increased latency only during concurrent FileZilla transfers

Goal: Identify the root cause of latency spikes that occur only when FileZilla handles multiple simultaneous transfers, even though the client maintains a connection pool to avoid repeated handshakes.

Uncertainty remains whether the delay originates from the client's internal queue management, from hitting the server's max‑connections‑per‑IP threshold, or from operating‑system socket buffer behavior under high concurrency.

Specific questions: Does FileZilla reuse pooled sockets across transfers in a way that creates head‑of‑line blocking? Does the queue algorithm introduce additional wait time when a transfer slot becomes free? Could OS TCP buffer settings exacerbate latency when many sockets are active simultaneously?