Flow control vs. Congestion control

In computer networks and architecture, reliable data transfer is of utmost importance for efficient communication between devices. To ensure this reliability in data transmission, two mechanisms are widely used, flow control and congestion control.

Both of these terms might seem alike at first, but both mechanisms have completely different purposes along with key differences. Understanding these key differences allows you to utilize these mechanisms in your computer network to ensure reliable and efficient data transmission.

Flow control

In the flow control mechanism, the receiver controls the sender by sending feedback. This ensures that the sender won't overflow the receiver's buffer by transmitting a large number of packets that the receiver would not be able to accommodate in its buffer.

Primary focus

Flow control primarily focuses on regulating the rate of transmission from the sender's side. This prevents the receiver from becoming overwhelmed by incoming data streams. Therefore, flow control helps avoid potential data loss of buffer overflow in the receiver.

Working

Flow control exchanges control signals, also known as feedback, between the sender and the receiver, and in turn, allows the sender to adjust its transmission rate according to the network's environment.

In short, the feedback helps maintain a balance between the sender's transmission speed and the receiver's buffer capacity.

Implementation

Flow control usually maintains the network by having the sender maintain a dynamic variable rwnd which indicates the receive window.

• The receiver "advertises" the rwnd value to the sender y sending feedback to it during acknowledgment.

• Sender sets its rwnd value equal to the advertised buffer space and limits the data packet according to the buffer space available.

• This ensures that the receiver buffer would never overflow, as the sender won't overwhelm the receiver's buffer by sending more data that can be accommodated at the receiver's side.

Congestion control

Congestion control mainly focuses on managing data traffic within the network, unlike flow control which manages data rates at the receiver's end.

The primary objective of congestion control is to prevent network congestion. Congestion happens when the data traffic within the network surpasses the network's available capacity, leading to inefficiency and performance degradation of the network.

Moreover, congestion can also lead to data loss due to multiple timeouts of unacknowledged packets.

Working principle

Congestion control mechanisms typically work by dynamically adjusting the data transmission rates based on the network's immediate conditions.

Various congestion control mechanisms are used like

• TCP Reno

• TCP New Reno

• TCP Cubic

All these mechanisms rely on detecting packet loss within the network and increasing or decreasing the transmission rates accordingly.

Implementation

Congestion control mechanisms use a dynamic function of the perceived network congestion called the cwnd or congestion window.

• The sender limits the transmission according to the cwnd variable. This helps the sender in avoiding traffic congestion in the computer network, thus avoiding potential data loss and network failure.

• The network sends feedback throughout the runtime and the cwnd variable is dynamically adjusted throughout the period to get efficient transmission rates without the risk of failures.

Summary

The differences between flow control and congestion control can be best illustrated by the following table: