the timeout is occurring as a result of TCP sender is receiving no ACKs in any respect.
abstract: sure, the timeout for the packet signifies that the ACK for this packet just isn’t obtained at some point of timeout, however, retransmissions normally occur after receiving dupacks, which normally occurs earlier than the timeout. Retransmitting one thing due to timeout is definitely fairly uncommon.
TCP receiver has to acknowledge each obtained section, irrespectively of whether or not that is an in-order or not in-order section. Within the latter case, TCP acknowledges final in-order obtained section. For the sender, because of this if some packets are literally delivered, some acks are obtained. Additionally, when TCP receives a number of dupacks (by default 3), it assumes that the section is misplaced and retransmits it. It additionally reacts to the congestion.
Beneath regular circumstances dupacks are obtained earlier than timer fires. And when timer fires that is normally becase no acks are obtained in any respect [*]
That is the distinction between simply down traces, the place TCP just isn’t going within the sluggish begin (e.g, after d1), and the traces after timeout. These are additionally known as minor and main congestion occasion respectively.
Minor occasion occurs after receiving 3 dupacks. Since acks are obtained community is delivering some packets, so it’s “believed to be not that congested”[**], and the response just isn’t that extreme. Main occasion occurs after a timeout. If there aren’t any ACKs obtained in any respect, TCP thinks that one thing important occurred, and begins from the start.
This one thing important might for instance occur if the trail of packets has modified and new path has a bottleneck with a lot decrease bandwidth.
- [*] tail losses are instance of non-normal circumstances
- [**] appologies for my consider concerning the foundations of congestion management algorithms, however i assume thath is what you must memorize