Why Does TCP Use a Sliding Window?

🎯 The Question

"Why does TCP use a Sliding Window protocol instead of simple Stop-and-Wait? How does it prevent a fast sender from overwhelming a slow receiver?"

---

⚡ 30-Second Elevator Pitch

In a naive Stop-and-Wait protocol, the sender sends 1 packet and waits for 1 ACK. Over high-latency networks, throughput collapses because the pipe sits empty for most of the Round Trip Time (RTT).

TCP Sliding Window solves this by enabling pipelining:

1. The receiver advertises its available buffer capacity via the Receive Window (rwnd) header field.
2. The sender is allowed to transmit up to rwnd bytes continuously without waiting for individual ACKs.
3. As the receiver processes data and sends ACKs, the window "slides" forward, keeping the network link saturated while guaranteeing the receiver's buffer never overflows.

---

🧠 Under-the-Hood: Flow Control in Action

---

🔬 Bandwidth-Delay Product (BDP)

The theoretical maximum throughput of a connection is defined by:

$\text{BDP} = \text{Bandwidth} \times \text{RTT}$

• Without Sliding Window: Throughput is bounded to $\frac{\text{Packet Size}}{\text{RTT}}$, wasting ~99% of fiber capacity.
• With Sliding Window: The window size matches BDP, keeping the network pipeline full at all times.

---

📌 Comparison Matrix: Stop-and-Wait vs. Sliding Window

Metric	Stop-and-Wait Protocol	TCP Sliding Window Protocol
Packets in Flight	Exactly 1 packet	Multiple packets (up to Window Size)
Throughput Utilization	Extremely low on high-latency links	Maximum (Saturates available bandwidth)
Receiver Protection	Natural (Slow, but safe)	Active Flow Control via advertised rwnd
Network Pipe Status	Mostly idle waiting for ACKs	Constantly full (Pipelined transmission)

---

💡 What Interviewers Ask Next (Follow-Up Traps)

1. "What happens if the receiver's application is stuck and advertises a Zero Window (rwnd = 0)?"

   • Answer: The sender pauses transmission. To avoid permanent deadlock (if the subsequent window update ACK is lost), the sender starts a Persist Timer and periodically sends 1-byte Zero-Window Probes to query the receiver's buffer status.

2. "What is the difference between Flow Control and Congestion Control in TCP?"

   • Answer: Flow Control (rwnd) protects the receiver's buffer from overflowing. Congestion Control (cwnd) protects the intermediate network routers from congestion collapses. The sender always uses $\min( ext{rwnd}, ext{cwnd})$.

---

Placement & Interview Takeaway

Interview Answer: TCP uses a sliding window to maximize throughput by pipelining multiple packets in flight. It simultaneously enforces receiver flow control through the advertised window (rwnd), ensuring high-speed data transmission without overflowing the receiver's memory buffer.

---

📺 Video Explanation