The Open Systems Interconnection (OSI) model—also called the OSI reference model—is a conceptual model that divides network communication and interoperability into seven abstract layers. It provides a standardized model that enables different applications, computer systems and networks to communicate.
The OSI model emerged as a solution to communication incompatibilities between the diverse array of networking protocols in use around the turn of the century. The layers of OSI gave developers and engineers a framework for building interoperable hardware and software across networks by providing a categorical approach to networking.
At each layer of the stack—typically shown in reverse order to illustrate how data moves through a network—the OSI model provides guidelines and criteria for network components and their unique computing functions.
The layers are:
The OSI model focuses on providing a list of tasks for engineers to complete in building each layer of a network architecture, instead of specifying protocols for communication between layers. Its theoretical approach enables developers to visualize and build highly complex computing networks, even without prior knowledge of the networking system itself. It also helps teams better understand how data traverses a network and tailor network functions with layer-specific coding.
Although the OSI model isn’t the direct basis for modern computer networking technologies, it’s had a profound impact on computing standards development, helping shape contemporary understandings of network architecture.
In the late 1970s and early 1980s, computer systems were becoming increasingly interconnected, but manufacturers often developed their own networking solutions, creating a patchwork of proprietary and non-interoperable systems.
Several early networking efforts attempted to address compatibility issues with the ARPANET (which laid the groundwork for the modern internet) and the TCP/IP protocol suite (commissioned by the Department of Defense). Both represented significant advancements, but they also highlighted the need for a more comprehensive and universally accepted approach.
Recognizing the growing importance of networking and the need for a universal framework, the International Organization for Standardization (ISO) and the International Telegraph and Telephone Consultative Committee (CCITT) initiated the development of a standardized networking model.
The ISO formally published the OSI model, a seminal framework for developing interoperable network solutions, in 1984. Unlike previous standardization attempts, the layered configuration of OSI enables disparate systems to communicate despite differences in their underlying architectures and protocols.
The OSI model remains integral to understanding network architecture even as technologies evolve and new models emerge. Whether a team is designing a simple local area network (LAN) or managing a complex global network, the principles of the OSI model provide a clear, structured approach to networking.
The OSI model includes seven distinct layers. The application layer (layer 7), the presentation layer (layer 6) and the session layer (layer 5) comprise the software layers of an OSI, where all transmissions to and from software apps (including operating systems and utilities, such as web browsers and email clients) occur.
The transport layer (layer 4) is the “heart of OSI,” handling all data communication between networks and systems. Finally, the network layer (layer 3), the data layer (layer 2) and the physical layer (layer 1) comprise the hardware layers of OSI, where data moves through the physical components of the network as it’s processed.
Data moves bi-directionally through the OSI model; each layer communicates with the layers below and above it in the stack. Furthermore, both the sending and receiving devices transmit data through the data layers; and senders and receivers often switch roles in the process.
For example, if a user wants to send an email to another person, the user would first write the email and send it. When the user presses “send,” their email goes to the application layer, which will choose the correct protocol (typically SMTP) and send the email to the presentation layer. The presentation layer then compresses the message data and sends it to the session layer, which initiates a communication session and sends the data to the transport layer for segmentation.
Since the email is going to another network, the email data must go to the network layer, where it’s divided into packets and then to the data link layer where it’s further broken down into frames. Those frames are subsequently transmitted through the physical layer (the recipient’s wifi), at which point the recipient’s device receives the bit stream and the email data traverses the same layers in reverse. At the end of the process, the email data lands in the application layer of the recipient’s device where it’s delivered, in human-readable form, to the recipient’s inbox.
The OSI model is foundational to protocol development, with each layer of the framework managing specific network processes.
The application layer is the OSI layer closest to the end user. It provides network services directly to user applications and facilitates communication between API endpoints and lower layers of the OSI model. In other words, software applications use the application layer to initiate communication with the network and send data to the presentation layer.
Applications themselves are not part of this layer. Rather, the application layer provides the protocols (HTTP, FTP, DNS and SMTP, for instance) that enable software to send and receive data. It’s responsible for processes such as:
Directory services.The application layer provides directory services—a shared database of information about network devices and users—to facilitate network resource management.
The presentation layer transforms data into a format that the application layer can accept for transmission across the network (from an EBCDIC-coded text file to an ASCII-coded file, for instance). Due to its role in converting data and graphics into a displayable format for the application layer, it is sometimes referred to as the syntax layer.
It supports secure sockets layer/transport layer security (SSL/TLS) protocols, JPEG protocols (for image compression) and MPEG protocols (for video The presentation layer transforms data into a format that the application layer can accept for transmission across the network (from an EBCDIC-coded text file to an ASCII-coded file, for instance). Due to its role in converting data and graphics into a displayable format for the application layer, it is sometimes referred to as the syntax layer.
It supports secure sockets layer/transport layer security (SSL/TLS) protocols, JPEG protocols (for image compression) and MPEG protocols (for video compression). The presentation layer is responsible for:
Data translation.The presentation layer converts data into the correct format (specified by the application layer) during the encapsulation process, as outgoing messages move down the protocol stack from sender to receiver.
Data compression.The presentation layer reduces the size of a data stream for transmissions and decompresses it for use.
Sometimes formatting and translation are reversed during the de-encapsulation process, as incoming messages move up the protocol stack. In those instances, outgoing messages are converted into the specified format during encapsulation, while incoming messages undergo a reverse conversion during de-encapsulation.
The session layer is responsible for session management, the process of establishing, managing and terminating connections—called "sessions"—between two or more computers. It initiates the connections between local and remote applications, keeping the session open long enough to transmit the necessary data and closing them when complete to preserve network resources.
Key functions of the session layer include:
Session interactions.The session layer manages user logon (establishment) and user logoff (termination), including any authentication protocols integrated into client software.
Session recovery.The session layer manages session failures and re-establishes connections if there are network problems.
It also establishes protocols for connecting and disconnecting sessions between related data streams, such as audio and video in web conferencing. Therefore, the session layer is often explicitly implemented in network environments that utilize remote procedure calls.
The transport layer uses protocols like transmission control protocol (TCP) and the user datagram protocol (UDP) to manage the end-to-end delivery of complete messages. It takes messages from the session layer and breaks them into smaller units (called “segments”), each with an associated header. At the destination, the transport layer reassembles the segments in the correct order to reconstruct the original message.
The transport layer also handles:
Service point addressing.The transport layer helps ensure that messages are delivered to the correct process by attaching a transport layer header (including a service point or port address).
Multiplexing.The transport layer allows multiple network applications to use the same connection simultaneously.
At the sender's end, the transport layer receives formatted data from the upper layers, performs segmentation and implements flow and error control to ensure accurate data transmission. It adds source and destination port numbers to the header and then forwards the segmented data to the network layer.
At the receiver's end, the transport layer reads the port number from the header and forwards the received data to the appropriate application. It also handles the sequencing and reassembly of the segmented data and retransmits data if errors are detected.
The transport layer provides two types of service.
With connection-oriented service, a three-part process including connection establishment, data transfer and termination (or disconnection), the data receiver sends an acknowledgment of receipt back to the sender when the data packet is delivered. Connectionless service, however, only involves data transfer. The receiver does not confirm receipt, which accelerates communication but can be less reliable than connection-oriented service.
The network layer of the OSI model is responsible for facilitating data transfer from one node to another across different networks. The network layer determines the best path (routing) for data to travel between nodes. If segments are too large, the network layer breaks them up into smaller “packets” for transport and reassembles them on the receiving end.
A network serves as a medium where multiple nodes (each with a unique address) can connect. The network layer allows nodes to send messages to nodes on other networks by providing the message content and the destination address, leaving the network to determine the optimal delivery path (which may involve routing through intermediate nodes).
The network layer primarily uses the Internet Protocol v4 (IPv4) and IPv6 and is responsible for:
Packet fragmentation and reassembly.The network layer splits large packets (those that exceed the size limits of the data link layer) into smaller ones for transmission and reassembles them at the destination.
Reliability isn’t guaranteed in the network layer; while many network layer protocols offer reliable message delivery, some do not. Furthermore, error reporting isn’t mandatory at this layer of OSI, so data senders may or may not receive confirmation of delivery.
The data link layer’s primary function is to manage error-free data transfer between multiple devices interacting on the same network.
The DLL is divided into two sublayers.
The logical link control (LLC) layer—which serves as an interface between the media access control (MAC) layer and the network layer—handles flow control, synchronization and multiplexing (where two or more data streams share a single connection to the host). The MAC layer controls how devices access network mediums and transmit data.
When the DLL receives a packet from the network layer, it divides the packet into data “frames”—according to the frame size of the network interface card (NIC)— and transmits it to the host using its MAC address.
DLL functions include:
Framing.The DLL allows the sender to transmit a set of bits (data) that are meaningful to the receiver by attaching special bit patterns to the beginning and end of the frame.
Error control.The DLL detects damaged or lost frames and manages retransmission (if necessary) to ensure data integrity.
Access control.When multiple devices share a single communication channel, the MAC sublayer determines which device has control over the channel at a given moment.
The physical layer comprises the physical network components responsible for transmitting raw data—in the form of “bits,” or strings of 1s and 0s—between devices (connectors, routers, repeaters and fiber optic cables, for instance) and a physical medium (like wi-fi).
The physical layer is responsible for:
Bit rate control.The physical layer defines the data transmission rates, often in bits per second.
Transmission mode.The physical layer defines how data will flow between connected devices (as simplex, half duplex or full duplex transmission).
The physical layer also defines how encoding occurs over a physical signal (using electrical voltage, radio or light pulses, for example).
The OSI reference model provides a theoretical underpinning that helps engineers and developers understand the intricacies of network communication. However, it’s sometimes compared with another networking model: the transmission control protocol/internet protocol (TCP/IP) model.
Unlike the OSI model, the TCP/IP model is based on standardized protocols that are widely and directly implemented and in real-world networks. It consists of four layers—instead of seven—but each layer corresponds to one or more layers of the OSI model.
Network access layer.Also called the data link layer or the physical layer, the network access layer of a TCP/IP network includes both the hardware and software components necessary for interfacing with the network medium, combining the OSI model’s physical and data link layers. It handles physical data transmission—by using Ethernet (for LANs) and ARP protocols—between devices on the same network.
Transport layer.The TCP/IP transport layer serves the same function as the OSI model's transport layer; it enables reliable data transfer between upper and lower layers. Using TCP and UDP protocols, it also provides mechanisms for error checking and flow control.
The OSI model's primary value lies in its educational utility and its role as a conceptual framework for designing new protocols, making sure that they can interoperate with existing systems and technologies.
However, the TCP/IP model's practical focus and real-world applicability have made it the backbone of modern networking. Its robust, scalable design and horizontal layering approach has driven the explosive growth of the internet, accommodating billions of devices and massive amounts of data traffic.
Its abstracted, vertically layered approach to networking enables modular protocol design, where each layer can be developed and updated independently.
The modularity of the OSI model encourages faster innovation in protocol development, since software engineers can integrate new technologies without overhauling the entire network stack.
It also enables developers to abstract away the lower layers of the model to simplify development.
Software engineers can separate the operating layers of each network component and organize them according to their primary roles in the network. This decomposability makes it easier for developers to conceptualize a network and share simplified models between development teams.
When a device on the network fails or an app loses connection, the OSI model allows teams to pinpoint and isolate the problematic layer to address any security issues or networking vulnerabilities without disrupting the entire framework.
Discover how a hybrid cloud infrastructure can power your AI strategy. Learn from IBM experts how to transform existing technology into an agile, AI-ready system, driving innovation and efficiency across your business operations.
Explore how hybrid cloud solutions can optimize your AI-driven business operations. Learn from case studies and featured solutions to see how companies are using IBM’s hybrid cloud to achieve greater efficiency, scalability and security.
Learn about the key differences between infrastructure as a service (IaaS), platform as a service (PaaS) and software as a service (SaaS). Explore how each cloud model provides varying levels of control, scalability and management to meet different business needs.
Discover the hidden costs of scaling generative AI and learn from experts how to make your AI investments more efficient and impactful.
Learn the fundamentals of IT management, including why it's critical for modern organizations and key features that ensure smooth, efficient operations across technology systems.
Discover a range of tutorials and resources to help you manage and support IT infrastructure, from server management to cloud integration, storage systems and network security.
Virtualize your storage environment and manage it efficiently across multiple platforms. IBM Storage Virtualization helps reduce complexity while optimizing resources.
Accelerate the impact of AI across the enterprise with a more intentional hybrid cloud.
Find the right cloud infrastructure solution for your business needs and scale resources on demand.