TCP Flow Control Mechanism Explained

TCP flow control is a key mechanism to prevent the sender from sending data too fast and causing the receiver's buffer overflow.

Sliding Window Mechanism

Window Size

• Receive Window (rwnd): Available buffer size advertised by the receiver
• Congestion Window (cwnd): Window size calculated by the sender based on network conditions
• Actual Send Window: min(rwnd, cwnd), taking the smaller of the two

Working Principle

1. Receiver Advertisement: Advertise current available receive window size in the window field of TCP segment
2. Sender Adjustment: Sender adjusts sending rate based on receiver's window size
3. Zero Window: When receiver buffer is full, advertise window size as 0
4. Window Probe: Sender periodically sends zero window probe segments to query if receiver window is available

Flow Control Process

Normal Situation

• When receiver has sufficient buffer, window size remains large
• Sender can continuously send multiple segments
• After receiver processes data, it updates window size and advertises to sender

Buffer Tension

• When receiver buffer is nearly full, reduce window size
• Sender correspondingly reduces sending rate
• Avoid data loss and retransmission

Zero Window Handling

• When receiver buffer is full, advertise window size as 0
• Sender stops sending data but maintains connection
• Periodically send zero window probe (1 byte data)
• After receiver has space, update window size

Timing of Window Updates

• After Receiving Data: Process data and release buffer space
• After Application Layer Reads Data: After reading data from receive buffer
• Zero Window Probe Response: Respond to sender's probe segment

Related Questions

• What is the difference between flow control and congestion control?
• Why do we need zero window probe mechanism?
• How does sliding window affect TCP throughput?