The OSI Model

The Ethernet, originally called the Alto Aloha network, was designed by the Xerox Palo Alto Research Center in 1973 to provide communication for research and development CP/M computers. When in 1976 Xerox started to develop the Ethernet as a 20Mbps product, the network prototype was called the Xerox Wire.

In 1980, when the Digital, Intel and Xerox standard was published to make it a LAN standard at 10 Mbps, Xerox Wire changed its name back to Ethernet. Ethernet became a commercial product in 1980 at 10 Mbps. The IEEE called its Ethernet 802.3 standard CSMA/CD (or carrier sense multiple access with collision detection).

As the 802.3 standard evolved, it has acquired such names as Thicknet (IEEE 10Base-5), Thinnet or Cheapernet (10Base-2), Twisted Ethernet (10Base-T) and Fast Ethernet (100Base-T). The design of Ethernet preceded the development of the seven-layer OSI model.

The Open System Interconnect (OSI) model was developed and published in 1982 by the International Organisation for Standardisation (ISO) as a generic model for data communication. The OSI model is useful because it is a broadly based document, widely available and often referenced.

Since modularity of communication functions is a key design criterion in the OSI model, vendors who adhere to the standards and guidelines of this model can supply Ethernet-compatible devices, alternative Ethernet channels, higherperformance Ethernet networks and bridging protocols that easily and reliably connect other types of data network to Ethernet.

Since the OSI model was developed after Ethernet and Signaling System #7 (SS7), there are obviously some discrepancies between these three protocols. Yet the functions and processes outlined in the OSI model were already in practice when Ethernet or SS7 was developed.

In fact, SS7 networks use point-to-point configurations between signalling points. Due to the point-to-point configurations and the nature of the transmissions, the simple data link layer does not require much complexity.

The OSI reference model specifies the seven layers of functionality. It defines the seven layers from the physical layer (which includes the network adapters), up to the application layer, where application programs can access network services.

However, the OSI model does not define the protocols that implement the functions at each layer. The OSI model is still important for compatibility, protocol independence and the future growth of network technology.

Implementations of the OSI model stipulate communication between layers on two processors and an interface for interlayer communication on one processor. Physical communication occurs only at layer 1. All other layers communicate downward (or upward) to lower (or higher) levels in steps through protocol stacks.

The following briefly describes the seven layers of the OSI model:

  • Physical layer

The physical layer provides the interface with physical media. The interface itself is a mechanical connection from the device to the physical medium used to transmit the digital bit stream.

The mechanical specifications do not specify the electrical characteristics of the interface, which will depend on the medium being used and the type of interface. This layer is responsible for converting the digital data into a bit stream for transmission over the network.

The physical layer includes the method of connection used between the network cable and the network adapter, as well as the basic communication stream of data bits over the network cable.

The physical layer is responsible for the conversion of the digital data into a bit stream for transmission when using a device such as a modem, and even light, as in fibre optics. For example, when using a modem, digital signals are converted into analogue audible tones which are then transmitted at varying frequencies over the telephone line.

The OSI model does not specify the medium, only the operative functionality for a standardised communication protocol. The transmission media layer specifies the physical medium used in constructing the network, including size, thickness and other characteristics.

  • Data link layer

The data link layer represents the basic communication link that exists between computers and is responsible for sending frames or packets of data without errors. The software in this layer manages transmissions, error acknowledgement and recovery.

The transceivers are mapped data units to data units to provide physical error detection and notification and link activation/deactivation of a logical communication connection. Error control refers to mechanisms to detect and correct errors that occur in the transmission of data frames.

Therefore, this layer includes error correction, so when a packet of data is received incorrectly, the data link layer makes system send the data again. The data link layer is also defined in the IEEE 802.2 logical link control specifications.

Data link control protocols are designed to satisfy a wide variety of data link requirements:

  • High-level Data Link Control (HDLC) developed by the International Organisation for Standardisation (ISO 3309, ISO 4335)
  • Advanced Data Communication Control Procedures (ADCCP) developed by the American National Standards Institute (ANSI X3.66)
  • Link Access Procedure, Balanced (LAP-B) adopted by the CCITT as part of its X.25 packet-switched network standard
  • Synchronous Data Link Control (SDLC) is not a standard, but is in widespread use. There is practically no difference between HDLC and ADCCP. Both LAP-B and SDLC are subsets of HDLC, but they include several additional features.
  • Network layer

The network layer is responsible for data transmission across networks. This layer handles the routing of data between computers. Routing requires some complex and crucial techniques for a packet-switched network design.

To accomplish the routing of packets sending from a source and delivering to a destination, a path or route through the network must be selected. This layer translates logical network addressing into physical addresses and manages issues such as frame fragmentation and traffic control.

The network layer examines the destination address and determines the link to be used to reach that destination. It is the borderline between hardware and software.

At this layer, protocol mechanisms activate data routing by providing network address resolution, flow control in terms of segmentation and blocking and collision control (Ethernet). The network layer also provides service selection, connection resets and expedited data transfers.

The Internet Protocol (IP) runs at this layer. The IP was originally designed simply to interconnect as many sites as possible without undue burdens on the type of hardware and software at different sites.

To address the shortcomings of the IP and to provide more a reliable service, the Transmission Control Protocol (TCP) is stacked on top of the IP to provide end-to-end service. This combination is known as TCP/IP and is used by most Internet sites today to provide a reliable service.

  • Transport layer

The transport layer is responsible for ensuring that messages are delivered error-free and in the correct sequence. This layer splits messages into smaller segments if necessary and provides network traffic control of messages.

Traffic control is a technique for ensuring that a source does not overwhelm a destination with data. When data is received, a certain amount of processing must take place before the buffer is clear and ready to receive more data.

In the absence of flow control, the receiver’s buffer may overflow while it is processing old data. The transport layer, therefore, controls data transfer and transmission.

This software is called Transmission Control Protocol (TCP), common on most Ethernet networks, or System Packet Exchange (SPE), a corresponding Novell specification for data exchange. Today most Internet sites use the TCP/IP protocol along with ICMP to provide a reliable service.

  • Session layer

The session layer controls the network connections between the computers in the network. The session layer recognises nodes on the LAN and sets up tables of source and destination addresses. It establishes a handshake for each session between different nodes.

Technically, this layer is responsible for session connection (i.e. for creating, terminating and maintaining network sessions), exception reporting, coordination of send/receive modes and data exchange.

  • Presentation layer

The presentation layer is responsible for the data format, which includes the task of hashing the data to reduce the number of bits (hash code) that will be transferred. This layer transfers information from the application software to the network session layer to the operating system.

The interface at this layer performs data transformations, data compression, data encryption, data formatting, syntax selection (i.e. ASCII, EBCDIC or other numeric or graphic formats), and device selection and control.

It actually translates data from the application layer into the format used when transmitting across the network. On the receiving end, this layer translates the data back into a format that the application layer can understand.

  • Application layer

The application layer is the highest layer defined in the OSI model and is responsible for providing user-layer applications and network management functions.

This layer supports identification of communicating partners, establishes authority to communicate, transfers information and applies privacy mechanisms and cost allocations.

It is usually a complex layer with a client/server, a distributed database, data replication and synchronisation. The application layer supports file services, print services, remote login and e-mail.

The application layer is the network system software that supports user-layer applications, such as word or data processing, CAD/CAM, document storage and retrieval and image scanning.