Understanding latency differences between UDP and TCP
This is an easy networking question that tests foundational knowledge of protocol trade-offs—the kind quant firms ask to ensure candidates understand why low-latency systems often choose UDP over TCP. It rewards clear thinking about the mechanisms that add latency to each protocol.
To answer questions like this, you need to understand the architectural differences between the two protocols: what guarantees each one provides, what computational and I/O overhead those guarantees impose, and how those overheads translate into wall-clock time. The question focuses on unicast scenarios, where both protocols are viable options.
- Connection setup and teardown costs
- Acknowledgement and retransmission logic
- In-order delivery and buffering requirements
- Checksum computation and error handling