Table of Contents
Retransmission Timeout (RTO) is a crucial parameter in TCP connections that determines how long a sender waits before retransmitting a packet. Proper calculation of RTO ensures stable and efficient data transmission, minimizing delays and avoiding unnecessary retransmissions.
Understanding RTO in TCP
RTO is dynamically calculated based on network conditions. It adapts to variations in round-trip time (RTT), which is the time it takes for a packet to travel from sender to receiver and back. Accurate RTO calculation helps prevent premature retransmissions or excessive waiting.
Calculating RTO
The calculation involves measuring the RTT and its variation over time. The standard algorithm uses the following formulas:
Estimated RTT is updated with each new measurement, and RTT deviation accounts for variability. The RTO is then set based on these values, often with a safety margin.
Key Steps in RTO Calculation
- Measure the RTT for each acknowledged packet.
- Update the Estimated RTT using an exponential weighted moving average (EWMA).
- Calculate the RTT deviation to account for network variability.
- Set the RTO as the sum of Estimated RTT and a multiple of the deviation.
Typical values for the multiple are 4 times the deviation, providing a balance between responsiveness and stability.