TCP Fast Retransmit

Understanding TCP Fast Retransmit in systems interviews

This question tests your grasp of TCP congestion control and reliability mechanisms—a topic that appears regularly in systems and networking interviews at firms building low-latency infrastructure. It asks you to explain a core algorithm that improves TCP performance by reducing the time spent waiting for timeouts.

To answer well, you should understand what triggers the algorithm, how it differs from the standard retransmission timeout approach, and why it matters for throughput. The question rewards clear reasoning about packet sequence numbers, acknowledgments, and the trade-off between responsiveness and false positives.

• Duplicate acknowledgments and their role as loss signals
• Retransmission timeout (RTO) vs. algorithmic loss detection
• Interaction with congestion control windows
• Real-world impact on latency-sensitive applications