TCP/IP Reference Model Explained

TCP/IP is a networking model. It divides the entire communication process into five layers: Application, Transport, Internet or Network, Data Link, and Physical. Each layer describes a specific part of the communication, along with the devices and protocols required to enable it. This tutorial explains the features and functions related to each layer.

Application Layer

The Application layer specifies the protocols and standards necessary for applications to connect to the network. For instance, the HTTP protocol defines how a web browser retrieves a web page from a web server.

The following figure illustrates the HTTP protocol in operation.

The HTTP protocol governs communication between the web browser and the web server. It establishes a standardized language that both entities use to exchange information. When a user enters a URL in a web browser's address bar and initiates a request, the HTTP protocol resolves the destination address, retrieves the requested content from the destination host, and displays it. If the content is unavailable, an error message is presented.

The Application layer does not define specific applications. It only establishes the standards, services, and protocols required for applications to connect to remote computers. Examples of application-layer protocols include HTTP, HTTPS, SNMP, NTP, SSH, FTP, TFTP, Telnet, DHCP, and DNS.

Transport layer

This layer is also known as the host-to-host layer. It provides a logical connection between two hosts. The main functions of this layer are the following:-

• Segmentation
• Reliability
• Flow control
• Connection multiplexing

Segmentation

Segmentation refers to the process of dividing a large data file into smaller, identifiable pieces, each known as a segment. For example, if a host needs to transmit a 100 MB file over a 1 Mbps connection, it cannot send the entire file at once. Instead, the file is divided into 100 or more segments, each sized to fit within the connection's capacity. This process is known as segmentation, and each resulting piece is called a segment. Because each segment is smaller than the connection's capacity, it can be transmitted efficiently. Upon receiving all segments, the destination host reassembles them into the original file. Segmentation enables hosts to transmit and receive data files of varying sizes over connections with different capacities.

Reliability

A connection is considered reliable if it provides the following services:-

• Recognize and resend the lost segments.
• Recognize the segment order and arrange segments in the correct order.
• Detect the duplicate segments and drop the extra segments.

The TCP protocol offers all of the aforementioned services. It assigns a unique sequence number to each segment, enabling the remote host to reconstruct the original order. After transmitting the segments, TCP waits for an acknowledgment from the remote host and retransmits any segments that do not receive an acknowledgment.

Certain applications, such as DHCP and DNS, do not require reliable connections. These applications typically use the UDP protocol for data transmission. Because UDP does not provide reliability, its segment headers omit the additional information required for reliable delivery, resulting in lighter headers. Segments with lighter headers traverse the network more quickly than those with heavier headers.

Flow control

Flow control mechanisms prevent the remote host from becoming overwhelmed by excessive segment transmission from the source host. Two primary types of flow control are ready/not-ready signals and windowing.

Ready/not-ready signals

In this method, the destination host transmits ready and not-ready signals to the source host, which adjusts its transmission accordingly. When the destination host's buffer is full, it sends a not-ready signal, prompting the source host to pause transmission. Once the destination host is prepared to receive additional segments, it sends a ready signal, and the source host resumes data transmission.

Windowing

In windowing, the source and destination hosts agree upon a window size, which specifies the number of segments the source host may send before requiring an acknowledgment. Upon receiving acknowledgment, the source host transmits the next set of segments.

Connection multiplexing

Multiplexing enables a host to establish multiple simultaneous connections with one or more hosts. Each connection, or session, may be reliable or unreliable. To differentiate sessions, the source host assigns unique source port numbers to each segment transmitted within a session.

Internet or Network Layer

There are two versions of the TCP/IP model. In the original model, this layer is called the Internet layer. In the updated version, it is called the Network layer. Its primary functions are addressing and routing. The IP protocol provides both functions.

Addressing

Each host in a network requires a unique address to communicate with other hosts. This unique address comprises a software address and a hardware address. The Internet (or Network) layer assigns software addresses, commonly known as IP addresses.

Routing

Analogous to the postal service delivering packages, routing services deliver data packets within a network. Routers perform this function by receiving data packets from source hosts and applying specific logic to filter, forward, and deliver them. A router examines the destination address of each incoming packet, identifies the destination host, and ensures delivery to that host.

Similar to how the post office filters and reroutes packages based on destination addresses, a router examines the destination address in a data packet and directs it along the most efficient path.

Link Layer (Data Link and Physical Layer)

This layer establishes standards and protocols for data transmission and physical connectivity. It also provides hardware addressing to identify devices on the local network. The primary functions of this layer are switching and device connectivity.

Switching, in general, is analogous to routing but uses hardware addresses rather than software addresses. Switches perform the switching function. In addition to switching, this layer defines standards and protocols for physical connectivity.

In the original TCP/IP model, both functions are included within the Link layer. In the updated model, these functions are separated: standards and functions directly related to physical transmission are assigned to the Physical layer. In contrast, those indirectly related are assigned to the Data Link layer.

In the updated model, switching is specified in the Data Link layer, while the standards and protocols for physical transmission are defined in the Physical layer.

Reference to other networking models

When the TCP/IP model was in development, many standards and protocols had already been defined in other networking models. Developers of the TCP/IP model, instead of defining these standards and protocols again, referenced them in the TCP/IP model with their respective model names, such as Ethernet standards and IEEE standards.

Whenever you see a standard mentioned with its model name, such as the Ethernet or IEEE standards in the TCP/IP model, consider it a referenced standard. A referenced standard or protocol is one originally defined in another networking model.

Conclusion

The TCP/IP model remains fundamental to modern networking, providing a robust framework for communication across diverse systems and technologies. It divides networking tasks into distinct layers (Application, Transport, Internet or Network, Data Link, and Physical) to enable interoperability, scalability, and efficient troubleshooting. Understanding the purpose and function of each layer is essential for designing, managing, and securing network infrastructures in a rapidly evolving technological landscape.