The OSI Model was created based on recommendations from the International Organization for Standardization (ISO) in 1980 and with the current standard published in 1996. The official title for the model is the ISO OSI (Open Systems Interconnection) Reference Model since it describes or relates to connecting systems that are open for communication with other systems. In the model, the functions of the communication system are standardized by categorizing them into abstract layers. The functions which are similar are grouped into the same layer and provide services to the layers above their existing layer.
What Does the OSI Model Do?
The OSI model depicts how data communications should take place. It splits the functions or processes into seven groups that are described as layers. When protocols or other standards are developed by other organizations such as the American National Standards Institute (ANSI), Institute of Electrical and Electronic Engineers (IEEE), and the International Telecommunications Union (ITU) formerly known as the CCITT (Comite Consultatif Internationale de Telegraphique et Telephone), they are placed into a layer of the model to help with communication protocol integration and conceptual understanding. The majority of major network and computer vendors, large commercial entities, and governments support the use of the OSI model. Each of the layers of the OSI model is intended to function with those above and below it respectfully within the model definition.
What Are the Seven Layers of the OSI Model?
Each of the layers of the OSI model has a numerical level or layer and plain text descriptor. The Seven Layers of the OSI Model are 1 – Physical, 2 – Data Link, 3 – Network, 4 – Transport, 5 – Session, 6 – Presentation, 7 – Application. A common mnemonic used to remember the OSI model layers starting with the seventh layer (Application) is: “All People Seem to Need Data Processing.” The lower two layers of the model are normally implemented through software and hardware solutions, while the upper five layers are typically implemented through the use of software only. The published “advantages” of the OSI Model are: enhanced learning/teaching, reduced intricacy, modular engineering, interoperable technology, accelerated advancement, and standard interface definitions. Unfortunately; however, the OSI Model has not been found to map well to real world networking implementations or issues as the technical world has evolved. It is the most recognized model; however, and is still often used to describe networking protocols, gear, problems, and solutions.
What Are the OSI Model Layer Functions?
Layer 7 – Application
The Application Layer’s function is to provide services to the end-user such as email, file transfers, terminal access, and network management. This is the primary layer of interaction for the end-user in the OSI Model.
Layer Six – Presentation
The Presentation Layer’s primary responsibility is to define the syntax that network hosts use to communicate. Compression and encryption fall in the functions of this layer. It is sometimes referred to as the “syntax” layer and is responsible for transforming information or data into format(s) the application layer can use.
Layer Five – Session
The Session Layer establishes process to process communications between two or more networked hosts. Under OSI, this layer is responsible for gracefully closing sessions (a property of TCP) and for session check pointing and recovery (not used in IP). It is used in applications that make use of remote procedure calls.
Layer Four – Transport
The Transport Layer is responsible for the delivery of messages between two or more networked hosts. It handles fragmentation and reassembly of messages and controls the reliability of a given link.
Layer Three – Network
The Network Layer is primarily responsible for establishing the paths used for data transfer on the network. Network routers operated at this layer which can commonly be divided into three sub-layers: Sub network access, Sub network-dependent convergence, and Sub network-independent convergence.
Layer Two – Data Link
The Data Link Layer is primarily responsible for communications between adjacent network nodes. Network switches and hubs operate at this layer which may also correct errors generated in the Physical Layer.
Layer One – Physical
The Physical Layer handles the bit level transmission between two or more network nodes. Components in this layer include connectors, cable types, pin-outs, and voltages which are defined by the applicable standards organization.
| Layer | Name |
|---|---|
| 7 | Application |
| 6 | Presentation |
| 5 | Session |
| 4 | Transport |
| 3 | Network |
| 2 | Data Link |
| 1 | Physical |
How Do Real World Protocols Map to the OSI Model?
The following are commonly used or implemented protocols mapped to the appropriate layer of the OSI Model (as best as they can be mapped). The problem with mapping well-known protocols to the OSI is that there is not a specific (or even general) agreement on how the protocols map to the model layers.
Layer Name Common Protocols
7 Application SSH, FTP, telnet
6 Presentation HTTP, SNMP, SMTP
5 Session RPC, Named Pipes, NETBIOS
4 Transport TCP, UDP
3 Network IP
2 Data Link Ethernet
1 Physical Cat-5
What is the TCP/IP Model?
The TCP/IP (Transmission Control Protocol / Internet Protocol) was created in the 1970s by DARPA. The model came from ARPANET and is also referred to as the “Internet Model” or less frequently as the “DoD Model.” The TCP/IP model defines four abstraction layers in RFC 1122 instead of seven which describe a general set of design guidelines and implementations of specific protocols for network communication. It provides end-to-end connectivity and addresses the formatting, addressing, transmittal, routing, and how to receive data. The four layer TCP/IP model is often compared to the OSI Reference Model. A major difference between the two definitions is that TCP/IP is descriptive while the OSI Reference Model was intended to be prescriptive. The related protocols and model itself for TCP/IP are maintained by the IETF (Internet Engineering Task Force).
TCP Model Principals
In RFC 1122, the TCP model emphasized the use of communication principles over the layering concept fundamental to the OSI. These principles include end-to-end and robustness. The original definition of the end-to-end principle assigned the maintenance of state and overall intelligence at the edges of the network and that the Internet would connect these edges while focusing on simplicity and speed. This has evolved; however, with the requirement for firewalls, web caching, network address translations, etc. and the principle continues to evolve with the modern realities. The robustness principle emphasizes conservation in sending information but liberal behavior in receiving.
TCP/IP Model Functions
The TCP/IP Model has four functions. Starting from the lowest level, these include the Physical Layer, the Link Layer, the Internet, and the transport layers.
Physical Layer – The Physical Layer consists of purely hardware and includes the network interface card, connection cable, satellite, etc.
Link Layer – Also referred to as the “Network Access Layer.” It is the networking scope of the local network connection that a host is attached. The lowest layer of IP, it is used to move data packets between the Internet Layer interfaces of two hosts on the same link. Controlling the process can be accomplished in either the software driver for the network card or via firmware in the chipset. The specifications for translating network addressing methods are included in the TCP/IP model, but lower level aspects are assumed to exist and not explicitly defined. A hierarchical encapsulation sequence is not dictated either.
Internet Layer – Handles the problem of sending data packets to or across one or more networks to a destination address in the routing process.
Transport Layer – The Transport Layer is responsible for end-end message transfer capabilities that are independent of the network. The specific tasks in this layer include error, flow, and congestion control, port numbers, and segmentation. Message transmission at this layer can either be connection-based as defined in TCP, or connectionless as implemented in the User Datagram Protocol (UDP).
The Internet Protocol performs two functions:
1 – Host identification and addressing. This function uses a hierarchical addressing system referred to as the IP address.
2 – Packet routing. This is the task of moving data packets from the source to destination host by sending the information to the next router or network node that is closer to the final destination. Information can be transported that relates to a number of upper layer protocols which are identified by a unique protocol number. Some examples are IGMP (Internet Group Management Protocol) and ICMP (Internet Control Message Protocol) that perform internetworking functions which help show the differences in the TCP/IP and OSI models.
How Do the OSI and TCP/IP Models Compare?
The upper or top three layers in the OSI Model (Application, Presentation, and Session Layers) are combined into a single layer only in the TCP/IP model in the Application layer. There are some OSI protocol applications which combine the three layers such as X.400, there is not a stated requirement for the TCP/IP protocol stack to implement a discreet structure above the Transport Layer. The Session Layer corresponds to the Telnet virtual terminal functionality that is part of text based protocols like SMTP and HTTP TCP/IP model Application Layer protocols. It also corresponds to the TCP and UDP port numbering system that is part of the Transport Layer in the TCP/IP model. There are some functions or applications that in the OSI Model are located in the Presentation Layer which is located in the Internet application layer that uses the MIME standard. This is used in Application Layer protocols such as SMTP and HTTP.
Due to IETF protocol development efforts not being concerned with explicit layering of the networking models, many of their protocols do not cleanly fit into the OSI Model. These issues have been cleaned up through the publishing of annexes to the original OSI Model which makes protocols such as IGML and ICMP defined as layer management protocols for the Network Layer. The IETF protocols can also be recursively encapsulated with tunneling protocols such as the GRE (Generic Routing Encapsulation). The baseline OSI documentation does not address the technology or possibility of tunneling, but it is addressed through extensions to the model such as with the transport layer gateways defined within the International Standardized Profile network. Due to the widespread adoption of TCP/IP protocols; however, most OSI development regarding extensions has been abandoned.
How Does the TCP/IP Model Map to Real World Networking?
The TCP/IP model has become the defacto standard for real world implementation of networking. Some of the real world protocol mappings to the TCP/IP Model layers are:
TCP/IP Model
Application Layer FTP, HTTP, POP3, IMAP, telnet, SMTP, DNS, TFTP
Transport Layer TCP, UDP, RTP
Internet Layer IP, ICMP, ARP, RARP
Network Interface Layer Ethernet, Token Ring, FDDI, X.25, Frame Relay, RS-232, v.35
TCP/IP Model Facts
Besides being more closely grounded in the reality of modern networking, there are several facts or differences regarding TCP/IP from the more academic OSI Model:
- TCP/IP was defined after the advent of the Internet. The OSI was defined prior to the Internet.
- Service interface and protocols are loosely defined.
- The protocol is loosely layered while the OSI defines strict layering.
- TCP supports reliable delivery of data packets while UDP supports connectionless communication that is not possible in the OSI. ISO requires that all data packets be reliably delivered.
- The biggest downside to the TCIP/IP model is that the more academics teach students to reference the OSI Model, the less they will learn about the TCP/IP model actually used in industry.
What is IPv6?
IPv6 (Internet Protocol 6) is the most recent version of the Internet Protocol based on the legacy IPv4 standard. Both IPv4 and IPv6 are demultiplexed at the media layer; however, IPv6 increases the IP address size to 128 bits (from 32 bits) to support significantly more layers of addressing hierarchy, a greater overall address space, and the new concept of “Scalability of Multicast addresses.” There is also a new type of Internet address introduced in the IPv6 standard called an “Anycast Address” which is used to send information to any number of a group of network nodes. IPv6 options are located in separate data packet headers located between the IPv6 and Transport Layer headers. These changes allow for more efficient forwarding of information on the network as well as greater flexibility for introducing new options in the future that may not even be thought of today.
Reading about why the OSI Model must die, go the way of Disco (isn’t that coming back in a limited degree?), and DivX:
Kill the Beast: Why the Seven-Layer Model Must Die.
No related posts.
Almost perfect.
awesome explanation.
I disagree with your statement that is must die. Having learned the OSI model when beginning Networking Classes I found it very helpful in learning good troubleshooting techniques. Yes the protocols may not line up perfectly but as a learning aid there hasn’t been a better quick reference tool that I have found then the OSI model.
which layer is Connection-less and connection-Oriented..?
swapna,
Based on Session layer protocol, transport layer will decide TCP or UDP should be used. connection oriented and connection less protcol comes under Transport layer.
hi
Super explanation on OSI model
Thanks.
HTTP is in Application layer not in Presentation layer!
the osi model is the concept/designed for all network and computer learning purposes. it’s true that technology changed. You have to embrace the new one but don’t forget the past.
Even though technology has changed, OSI model is preferred because of uniformity and easy to understand
A few picky items. Under What is the TCP/IP model? you said that TCP/IP was descriptive, OSI prescriptive. Effectively you have that backwards. Technically both are descriptive, but the RFCs prescribe how and what you must do to use the TCP/IP suite on the internet. The OSI/RM has always been descriptive (by definition – that’s what Reference Model means) of how networking works, and has never prescribed what you need to do. They are for different purposes.
Also, you said that the OSI was before the internet, TCP/IP after. Again, backwards. TCP/IP, in the 70s, was used by the DOD & their collaborating universities to create the DARPANET (later the ARPANET) [and TCP/IP was created bit by bit as needed to get those communications to work -- prescriptive]. The OSI was written in part to help clarify all the issues that the inadequacies of the TCP/IP had in relating to other (often proprietary) systems [in other words, afterwards]. So the OSI/RM is JUST a reference model that can be used by anybody, regardless of their actual networking system.
This does not imply that one is better than the other. The OSI was designed to describe what is involved in networking so that new networking systems could be designed already knowing what would be involved. TCP/IP was created to network specific computers together, and modified as it turned out not to be generic enough for all the systems out there. Now it drives the largest [computer] network in existence. Many of the RFCs are deprecated because of specific problems, or because someone academic studied the OSI and realized potential problems. TCP/IP by nature must change constantly, the OSI might occasionally require tweaks, but was designed to be universal. Each is needed, the OSI to understand HOW networking works, the TCP/IP to be sure that one network (the Internet) works properly. If you want to hook into the internet, you do not need TCP/IP internally, but if not, you’ll need gateways. So it’s easier to design a TCP/IP network internally & you’re good to hook up to the internet when you get permission (IPs, domain name, etc).