The Open Systems Interconnection (OSI) networking model defines a conceptual framework for communications between computer systems. The model is an ISO standard which identifies seven fundamental networking layers, from the physical hardware up to high-level software applications.
Each layer in the model handles a specific networking function. The standard helps administrators to visualize networks, isolate problems, and understand the use cases for new technologies. Many network equipment vendors advertise the OSI layer that their products are designed to slot into.
OSI was adopted as an international standard in 1984. It remains relevant today despite the changes to network implementation that have occurred since first publication. Cloud, edge, and IoT can all be accommodated within the model.
In this article, we’ll explain each of the seven OSI layers in turn. We’ll start from the lowest level, labelled as Layer 1.
1. Physical Layer
All networking begins with physical equipment. This layer encapsulates the hardware involved in the communications, such as switches and cables. Data is transferred as a stream of binary digits – 0 or 1 – that the hardware prepares from input it’s been fed. The physical layer specifies the electrical signals that are used to encode the data over the wire, such as a 5-volt pulse to indicate a binary “1.”
Errors in the physical layer tend to result in data not being transferred at all. There could be a break in the connection due to a missing plug or incorrect power supply. Problems can also arise when two components disagree on the physical encoding of data values. In the case of wireless connections, a weak signal can lead to bit loss during transmission.
2. Data Link Layer
The model’s second layer concerns communication between two devices that are directly connected to each other in the same network. It’s responsible for establishing a link that allows data to be exchanged using an agreed protocol. Many network switches operate at Layer 2.
The data link layer will eventually pass bits to the physical layer. As it sits above the hardware, the data link layer can perform basic error detection and correction in response to physical transfer issues. There are two sub-layers that define these responsibilities: Logical Link Control (LLC) that handles frame synchronization and error detection, and Media Access Control (MAC) which uses MAC addresses to constrain how devices acquire permission to transfer data.
3. Network Layer
The network layer is the first level to support data transfer between two separately maintained networks. It’s redundant in situations where all your devices exist on the same network.
Data that comes to the network layer from higher levels is first broken up into packets suitable for transmission. Packets received from the remote network in response are reassembled into usable data.
The network layer is where several important protocols are first encountered. These include IP (for determining the path to a destination), ICMP, routing, and virtual LAN. Together these mechanisms facilitate inter-network communications with a familiar degree of usability. However operations at this level aren’t necessarily reliable: messages aren’t required to succeed and may not necessarily be retried.
4. Transport Layer
The transport layer provides higher-level abstractions for coordinating data transfers between devices. Transport controllers determine where data will be sent and the rate it should be transferred at.
Layer 4 is where TCP and UDP are implemented, providing the port numbers that allow devices to expose multiple communication channels. Load balancing is often situated at Layer 4 as a result, allowing traffic to be routed between ports on a target device.
Transport mechanisms are expected to guarantee successful communication. Stringent error controls are applied to recover from packet loss and retry failed transfers. Flow control is enforced so the sender doesn’t overwhelm the remote device by sending data more quickly than the available bandwidth permits.
5. Session Layer
Layer 5 creates ongoing communication sessions between two devices. Sessions are used to negotiate new connections, agree on their duration, and gracefully close down the connection once the data exchange is complete. This layer ensures that sessions remain open long enough to transfer all the data that’s being sent.
Checkpoint control is another responsibility that’s held by Layer 5. Sessions can define checkpoints to facilitate progress updates and resumable transmissions. A new checkpoint could be set every few megabytes for a file upload, allowing the sender to continue from a particular point if the transfer gets interrupted.
Many significant protocols operate at Layer 5 including authentication and logon technologies such as LDAP and NetBIOS. These establish semi-permanent communication channels for managing an end user session on a specific device.
6. Presentation Layer
The presentation layer handles preparation of data for the application layer that comes next in the model. After data has made it up from the hardware, through the data link, and across the transport, it’s almost ready to be consumed by high-level components. The presentation layer completes the process by performing any formatting tasks that may be required.
Decryption, decoding, and decompression are three common operations found at this level. The presentation layer processes received data into formats that can be eventually utilized by a client application. Similarly, outward-bound data is reformatted into compressed and encrypted structures that are suitable for network transmission.
TLS is one major technology that’s part of the presentation layer. Certificate verification and data decryption is handled before requests reach the network client, allowing information to be consumed with confidence that it’s authentic.
7. Application Layer
The application layer is the top of the stack. It represents the functionality that’s perceived by network end users. Applications in the OSI model provide a convenient end-to-end interface to facilitate complete data transfers, without making you think about hardware, data links, sessions, and compression.
Despite its name, this layer doesn’t relate to client-side software such as your web browser or email client. An application in OSI terms is a protocol that caters for the complete communication of complex data through layers 1-6.
HTTP, FTP, DHCP, DNS, and SSH all exist at the application layer. These are high-level mechanisms which permit direct transfers of user data between an origin device and a remote server. You only need minimal knowledge of the workings of the other layers.
The seven OSI layers describe the transfer of data through computer networks. Understanding the functions and responsibilities of each layer can help you identify the source of problems and assess the intended use case for new components.
OSI is an abstract model that doesn’t directly map to the specific networking implementations commonly used today. As an example, the TCP/IP protocol works on its own simpler system of four layers: Network Access, Internet, Transport, and Application. These abstract and absorb the equivalent OSI layers: the application layer spans OSI L5 to L7, while L1 and L2 are combined in TCP/IP’s concept of Network Access.
OSI remains applicable despite its lack of direct real-world application. It’s been around so long that it’s widely understood among administrators from all backgrounds. Its relatively high level of abstraction has also ensured it’s remained relevant in the face of new networking paradigms, many of which have targeted Layer 3 and above. An awareness of the seven layers and their responsibilities can still help you appreciate the flow of data through a network while uncovering integration opportunities for new components.