Why TCP Uses a Sliding Window to Keep High-Latency Links Busy

Stop-and-Wait transmission performs poorly on high-latency networks because the link sits idle for most of each round trip. The bandwidth-delay product (bandwidth multiplied by RTT) measures how much data can be in flight at once. For example, a 10 Gbps transatlantic link with a 100 ms RTT has a BDP of about 125 MB, so Stop-and-Wait would leave almost all of that capacity unused. TCP's sliding window addresses this by letting a sender keep multiple segments in flight while tracking which data has been acknowledged. The article illustrates this with an interactive visualizer that divides the window into acknowledged, in-flight, usable, and blocked zones.