Updated 1 Nov 2026 · 7 min read
On this page
After reading this you will be able to name the seven OSI layers in order, say what each one is responsible for according to the standard itself, match common protocols and devices to a layer, translate between the OSI and TCP/IP models, and use the layers as a checklist when something on a network stops working. You will also see which placements the standards leave open, so you can answer carefully when a question touches them.
Where the model comes from
The OSI reference model is published as ITU-T Recommendation X.200 (07/94), and the same text is published by ISO and IEC as ISO/IEC 7498-1. The ITU still lists X.200 (07/94) as in force. It defines seven layers, numbered from the bottom: Physical (1), Data Link (2), Network (3), Transport (4), Session (5), Presentation (6) and Application (7).
The key idea is layering. Each layer provides a service to the layer above it, using the services of the layer below. Peers at the same layer on two systems talk using that layer's protocol, and X.200 calls the unit of data a layer exchanges its protocol data unit (PDU): protocol control information, such as a header, plus possibly user data from the layer above.
The seven layers, top to bottom
| Layer | What X.200 says it does | Examples |
|---|---|---|
| 7 Application | The only way an application process reaches the OSI environment. It has no layer above it. | Telnet, FTP, SMTP, DNS, SNMP (RFC 1122 lists these as application layer protocols) |
| 6 Presentation | Represents the information applications exchange, giving them syntax independence, and selects mutually acceptable transfer syntaxes. | No separate layer in the Internet suite (see mapping below) |
| 5 Session | Lets cooperating presentation entities organise and synchronise their dialogue and manage their data exchange. | No separate layer in the Internet suite |
| 4 Transport | Transparent, end-to-end transfer of data. It is not concerned with routing and relaying. | TCP, UDP |
| 3 Network | Transfer between transport entities across the network, handling routing and relaying so that the Transport Layer does not have to. Any relaying happens here or below. | IP, ICMP, IGMP. Routers. |
| 2 Data Link | Carries data between network entities over one or more physical connections, and detects and possibly corrects Physical Layer errors. | Ethernet and IEEE 802 encapsulation. Bridges and Layer 2 switches. |
| 1 Physical | The mechanical, electrical, functional and procedural means to activate, maintain and deactivate physical connections for bit transmission. | Cabling, connectors, repeaters |
A few of those device placements come from the IETF rather than X.200. RFC 1812 describes routers as devices that perform the network layer forwarding function, and says packet switching devices that operate at the Link Layer are usually called bridges. It also treats segments joined by bridges or repeaters as a single physical network, because those devices are transparent to IP. The repeater placement follows X.200 itself, which allows a physical connection to pass through intermediate systems that each relay bit transmission within the Physical Layer. The CCNA blueprint distinguishes Layer 2 and Layer 3 switches.
PDU names at each layer
Every layer has a PDU, but the everyday names come from the Internet protocols, and RFC 1122 defines them:
- Segment: the unit of end-to-end transmission in TCP, a TCP header followed by application data.
- Datagram: the IP datagram, an IP header followed by transport layer data. UDP also calls its unit a user datagram.
- Packet: the unit passed between the internet layer and the link layer, an IP header and data. A packet may be a whole datagram or a fragment of one.
- Frame: the unit of transmission in a link layer protocol, a link-layer header followed by a packet.
- Bits: the Physical Layer's job, in X.200's words, is bit transmission.
Encapsulation follows from those definitions. Application data gets a TCP header and becomes a segment. The segment is carried inside an IP datagram. Each packet, which is the whole datagram or a fragment of it, then travels inside a link-layer frame, which the Physical Layer sends as bits.
Mapping OSI to the TCP/IP model
RFC 1122 describes the Internet protocol suite in four layers: Application, Transport, Internet and Link. It says directly that the Internet application layer combines the functions of the top two OSI layers, Presentation and Application. RFC 1812 adds that the Link Layer contains everything below the Internet Layer and above the Physical Layer, and warns that some older Internet documents call the link layer the Network Layer, which is not the same as the OSI Network Layer.
| OSI layer | Internet suite layer (RFC 1122) |
|---|---|
| 7 Application, 6 Presentation | Application |
| 5 Session | Not mentioned in RFC 1122 |
| 4 Transport | Transport |
| 3 Network | Internet |
| 2 Data Link | Link |
| 1 Physical | Below the Link Layer (RFC 1812) |
RFC 1122 does not mention the OSI Session Layer, so any mapping of layer 5 onto the Internet suite is an interpretation, not something the standard states.
Placements that are debated
Some protocols do not sit cleanly in one OSI layer, and the standards that define them do not assign an OSI layer number. Treat these as debated rather than learning one answer as fact.
ARP. RFC 826 describes ARP as a way to convert protocol addresses such as IP addresses into local network addresses such as 48-bit Ethernet addresses, and its packets travel in Ethernet frames with their own protocol type. RFC 1122 covers ARP inside its Link Layer chapter. It serves the network layer but is carried directly in link-layer frames, so its OSI layer is a matter of interpretation rather than definition.
TLS. RFC 9846, the current TLS 1.3 specification (it obsoleted RFC 8446 in July 2026), says TLS needs only a reliable, in-order data stream from the transport underneath, and that higher-level protocols can layer on top of TLS transparently. So TLS sits above a transport such as TCP and below an application protocol. Neither RFC gives it an OSI layer number, and its name (Transport Layer Security) does not settle the question either.
ICMP. RFC 1122 calls ICMP an integral part of IP, although it is architecturally layered on top of IP and uses IP to carry its data just as TCP or UDP does.
Troubleshooting from the bottom up
The layers give you an order to check things in. Working from layer 1 upwards means you never waste time on an application setting when the cable is unplugged. These exam-style scenarios show the reasoning:
| Symptom | Layer to check first | Why |
|---|---|---|
| No link light on the switch port, interface down | 1 Physical | No physical connection means no bit transmission |
| Interface is up but its error counters keep climbing | 1 and 2 | X.200 gives the Data Link Layer the job of detecting errors that occur in the Physical Layer |
| Host can reach local devices but nothing on other subnets | 3 Network | Check the IP address, subnet mask and default gateway |
| Ping to the server works but the service port does not answer | 4 Transport | IP reachability is proven, so look at the TCP or UDP port |
| Ping to an IP address works, the same host by name fails | 7 Application | DNS is an application layer support protocol in RFC 1122 |
CompTIA's Network+ objectives list incorrect gateway, IP address or subnet mask among network services issues, and list cabling and physical interface issues as a separate group.
How this fits your exam
CompTIA Network+ N10-009. The Networking concepts domain (23%) in CompTIA's objectives summary lists "OSI model layers: physical, data link, network, transport, session, presentation, application." The Network troubleshooting domain (24%) covers the troubleshooting methodology and cabling and physical interface issues. The Network+ N10-009 study guide walks through all five domains.
Cisco 200-301 CCNA. The CCNA Exam v1.1 blueprint does not list the OSI model as a topic by name, but it uses layer numbers throughout: 1.1.b Layer 2 and Layer 3 switches, 2.3 Layer 2 discovery protocols, 2.4 Layer 2/Layer 3 EtherChannel and 5.7 Layer 2 security features. It also lists 1.5 "Compare TCP to UDP". See the CCNA 200-301 guide.
CompTIA A+ 220-1201. CompTIA's objectives summary for A+ Core 1 (V15) does not name the OSI model. Its Networking domain (23%) covers protocols and ports, SOHO networks including IP addressing, and networking tools, so the layer vocabulary above is useful background there. The A+ 220-1201 guide has the full outline.
Frequently asked questions
Is the OSI model still a current standard?
Yes. The ITU lists Recommendation X.200 (07/94) with the status "In force", and its foreword states that the identical text is published as ISO/IEC 7498-1. The Internet itself is built on the TCP/IP suite described in RFC 1122, which is why you learn both.
What is the difference between a PDU and an SDU?
In X.200, an (N)-protocol-data-unit is what peer entities at layer N exchange: protocol control information plus possibly user data. An (N)-service-data-unit is data from the layer above whose identity is preserved when it is transferred between the peer entities of that higher layer, and which layer N does not interpret. In short, the SDU is the payload and the PDU is the payload plus that layer's control information.
Why does the TCP/IP model have fewer layers?
RFC 1122 folds the OSI Presentation and Application layers into a single application layer, and it does not mention a Session Layer. At the bottom, it has a single link layer, with RFC 1812 placing the physical medium below it.
Sources
- X.200: Information technology - Open Systems Interconnection - Basic Reference Model: The basic model
- RFC 1122: Requirements for Internet Hosts - Communication Layers
- RFC 1812: Requirements for IP Version 4 Routers
- RFC 9846: The Transport Layer Security (TLS) Protocol Version 1.3
- CompTIA Network+
- CCNA Exam v1.1 (200-301)
Put the layers to work with the free Network+ N10-009 practice questions, and use the Network+ exam guide to plan the rest of your study.