Flow Control in the Transport Layer

Introduction

One of the important responsibilities of the Transport Layer is to ensure that the sender does not transmit data faster than the receiver can process it. This mechanism is known as Flow Control.

Flow control maintains a balance between the rate at which data is produced by the sender and the rate at which it is consumed by the receiver. If this balance is not maintained, data may be lost or the communication system may become inefficient.

Definition

Flow Control is a mechanism that regulates the rate of data transmission between the sender and the receiver so that the receiver is not overwhelmed with more data than it can handle.

Its primary objective is to prevent buffer overflow and data loss at the receiving side.

Why is Flow Control Needed?

Consider two situations.

Case 1: Sender is Faster than Receiver

Suppose,

Sender transmits 1000 packets/second
Receiver can process only 500 packets/second


Sender  ---------------------> Receiver

1000 Packets/sec              Can process only
                              500 Packets/sec

The receiver's memory (buffer) becomes full.

Result:

Buffer overflow
Packet loss
Retransmissions
Reduced performance

Case 2: Receiver is Faster than Sender

Suppose,

Sender transmits 100 packets/second
Receiver can process 1000 packets/second


Sender -----------------------> Receiver

100 Packets/sec               Can process
                              1000 Packets/sec

Now the receiver remains idle waiting for data.

Result:

Poor utilization
Reduced efficiency

Therefore, there must be a balance between the sender's transmission speed and the receiver's processing speed.

Producer and Consumer Concept

Flow control can be understood using the Producer–Consumer model.

Producer → Creates data.
Consumer → Receives and processes data.


Producer  -----------------> Consumer

Produces data               Consumes data

The producer should not produce data faster than the consumer can consume it.

Pushing and Pulling

Data can be delivered in two different ways.

1. Pushing

In Pushing, the producer sends data without waiting for a request from the consumer.


Producer
    │
    │ Sends data automatically
    ▼
Consumer

Characteristics

Sender controls communication.
Data is sent continuously.
Receiver may become overloaded.
Flow control is required.

Example

A video camera continuously sends video frames.

If the receiver becomes slow,

Frames accumulate.
Buffer becomes full.
Some frames may be lost.

2. Pulling

In Pulling, the consumer requests data only when it is ready.


Consumer
     │
     │ Request
     ▼
Producer
     │
     │ Sends data
     ▼
Consumer

Characteristics

Receiver controls communication.
Producer sends data only after request.
No receiver overload.
Flow control is generally unnecessary.

Example

Downloading a webpage.

The browser requests data, and the server sends only the requested information.

Pushing vs Pulling

Feature	Pushing	Pulling
Who initiates communication?	Producer	Consumer
Data sent	Automatically	On request
Receiver overload	Possible	Not possible
Need for Flow Control	Yes	Usually No

Flow Control at the Transport Layer

Communication at the transport layer involves four entities.


Application Process (Sender)

        │

Sender Transport Layer

        │

Receiver Transport Layer

        │

Application Process (Receiver)

Each entity has a different role.

1. Sending Application Process

Acts as a Producer.

Generates data.
Pushes message chunks to the transport layer.


Application

Produces Message

↓

Transport Layer

2. Sending Transport Layer

Acts as both

Consumer
Producer

Receives data from the application.
Encapsulates the data into packets.
Sends packets to the receiving transport layer.


Application

↓

Sender Transport Layer

Receives Data
↓

Creates Packet
↓

Sends Packet

3. Receiving Transport Layer

Also performs two roles.

Receives packets.
Removes the transport header.
Produces application messages.


Packet

↓

Receiver Transport Layer

↓

Original Message

4. Receiving Application Process

Acts as a Consumer.

Unlike the sender, it does not receive data continuously.

Instead,

it requests data whenever it is ready.

Thus, the last delivery follows the Pulling model.

Where is Flow Control Required?

According to the transport-layer communication model, flow control is required in two places.

Case 1

Between

Sending Application Layer

and

Sending Transport Layer


Application

↓

Transport Layer

If the transport layer buffer becomes full,

it tells the application


STOP SENDING

When space becomes available,


START SENDING AGAIN

Case 2

Between

Receiving Transport Layer

and

Sending Transport Layer


Sender Transport Layer

↓

Receiver Transport Layer

If the receiver buffer becomes full,

the receiver sends a signal


STOP

When space becomes available,


CONTINUE

Thus, data loss is prevented.

Role of Buffers

Flow control is usually implemented using buffers.

What is a Buffer?

A buffer is a temporary memory area used to store packets before they are processed.

Buffers exist at

Sender Transport Layer
Receiver Transport Layer

Sender Buffer


Application

↓

Sender Buffer

↓

Network

Purpose

Temporarily stores outgoing packets.
Prevents packet loss if the network is busy.

If full,


Application

STOP Sending

Receiver Buffer


Network

↓

Receiver Buffer

↓

Application

Purpose

Stores incoming packets until the application processes them.

If full,

Receiver informs the sender


STOP Sending

After processing some packets,


CONTINUE Sending

Flow Control Using Buffers

Step 1

Application generates data.


Application

↓

Sender Buffer

Step 2

Sender sends packets.


Sender Buffer

↓

Network

Step 3

Receiver stores packets.


Network

↓

Receiver Buffer

Step 4

Application reads packets.


Receiver Buffer

↓

Application

Step 5

If receiver buffer becomes full,


Receiver Buffer

FULL

↓

Signal

STOP

Step 6

Sender temporarily stops transmission.


Sender

Waits

Step 7

After the application processes some packets,

the receiver buffer has free space.


Receiver

↓

Buffer Empty

↓

Signal

CONTINUE

Sender resumes transmission.

Complete Flow Control Diagram

Advantages of Flow Control

Prevents buffer overflow at the receiver.
Avoids packet loss.
Matches the sender's speed with the receiver's processing capability.
Improves communication efficiency.
Reduces unnecessary retransmissions.
Ensures reliable end-to-end communication.

Real-Life Analogy

Imagine a supermarket checkout counter.

Customers are the data packets.
Cashier is the receiver.
Queue area is the buffer.

If customers arrive faster than the cashier can serve them, the queue becomes full. The security guard temporarily stops new customers from entering until there is space again.

Similarly, in a network:

The sender produces packets.
The receiver processes them.
The buffer temporarily stores packets.
Flow control prevents the sender from transmitting more data when the receiver's buffer is full.

Summary

Flow Control is a transport-layer mechanism that regulates the rate of data transmission so that the sender does not overwhelm the receiver.
Data transfer can occur through pushing, where the producer sends data without waiting for requests (requiring flow control), or pulling, where the consumer requests data only when ready (typically eliminating the need for flow control).
At the transport layer, the sending application produces data, the sending transport layer encapsulates and forwards it, the receiving transport layer decapsulates it, and the receiving application consumes it.
Flow control is commonly implemented using buffers at both the sender and receiver. When a buffer becomes full, the receiving side signals the sender to pause transmission, and when space becomes available, it signals the sender to resume, thereby preventing data loss and ensuring efficient communication.