Round Trip Time (RTT) and Time Out

The  size and  the complexity  of computer  networks  have grown  in past years. To achieve  an efficient  and reliable transmission  some  protocols inevitably need to handle  complicated network  traffics and unexpected transmission losses. These problems  usually  are  referred to as flow  control  and congestion control. The technologies  of managing complex computer  network  need to be more  circumspect not only  in the sending  host and  receiving  hosts but also  the intermediate routes. One of these protocols  is the  transmission control  protocol that has a responsibility  of ensuring reliability.

Because TCP  guarantees the reliable delivery  of data it retransmits  each segment  if an ACK is not received in a certain period of time. TCP  sets this  timeout as a function of the round trip time it expects between  the two ends  of eh connection . unfortunately given  the range of possible RTT between  any pair  of hosts  in the internet as well  as the variation in RTT between  the same two  hosts  over time. Choosing  appropriate timeout  value is  not very easy. To address this problem TCP uses an adapt retransmission mechanism.

Round  trip  time (RTT)  is an important in determining the behaviour of  a TCP  connection  passively estimating  RTT  is useful  in measuring  the congestion window size and retransmission  timeout of a connection as well as the  available bandwidth on a path. This information can help  determine factors that limit data flow rates  and can congestion. When  known  at  a network  link along the path RTT can also  aid efficient queen management  and buffer provisioning. Additionally  RTT can be used  to improve node  distribution  in peer to peer  and overlay  networks.

 The  amount  of time  set at  the time is however of great importance. If  the time  expires too quickly then premature time outs will be  generated during  usual  case and thus unnecessary retransmission will occur. On the  other hand  if a long time  is set for the timer the TCP will slowly respond to the segment  loss which  means longer delays in the  transportation and overall  end to end  delay. Therefore  the TCP time out must be set to  be as optimum  as possible.