Make your own free website on

CET1173C Lecture #3 - OSI Model Layer 2: The Datalink Layer

Lecture Only
The general definition and functional description of the DataLink layer
The concepts of Protocols and Packets, including,
The specifications of the major Ethernet protocols,
The Ethernet frame,
The relationship between the physical Layer and the DataLink layer,
The functions of the two major sublayers of the DataLink layer: The MAC and the LLC.
The student will become familiar with the DataLink layer protocols, technologies and devices and be able to recognize and describe the various major DataLink layer technologies with an emphasis on Ethernet LAN design and installation considerations concerning the usage and performance of the network.


  1. This is a good time to mention Service Access Points. The hardware card driver communicates with the DataLink layer through a "well known" memory address which the two layers are designed to find each other at. Each posts requests, reads the other layer's status messages and passes actual data back and forth through these addresses. Each layer must attach to the SAP (not to be confused with the Server Advertising Protocol used by Novell servers over either TCP/IP or IPX/SPX) of the layer immediately below it and the OSI layer driver immediately above it.

  2. In strict OSI theory, a particular layer driver numbered N can only communicate information heading out of the system to layer N – 1, and can only move data arriving at this system up to layer N + 1. It is interesting to note that each layer N forms packets for transmission to a remote system that are addressed and intended for that remote system's layer N. For example, a file is to be copied from PC #1 to PC #2. The DataLink layer of PC #1 will form an Ethernet packet that includes the destination MAC of PC #2. So the Layer 2 (DataLink) of PC #1 is creating a packet for the Layer 2 driver of PC #2 to get and handle. No other PC's Layer 2 would be able to process the packet since the MAC would be incorrect. Similarly, the Layer 3 (Network) packet encapsulated in the layer 2 packet is intended for PC #2's layer 3 driver directly. Only PC #2's layer 3 driver will be able to recognize and respond to the packet.

  3. Likewise the Layer 4 (Transport) packet that the Layer 3 IP packet delivers is a communication directly from the Layer 4 driver of PC #1 to the Layer 4 driver of PC #2. So the peer layers are in direct communication with each other. This is facilitated by the lower layers of the network and they are facilitating the upper layers of the network to the point where the information exchange exits the network stack at the Application layer interfaces to the operating system and applications themselves. This means that for any layer N of a given node, it establishes communications with: Local Layer N + 1 and Local Layer N - 1 and Foreign node Layer N.

  4. Here the SAP's appear as the small vertical arrows between the layers of the TCP/IP stack in RAM. Each layer communicates directly with the layer directly above and below it. The layer forms a packet that will be encapsulated by the packet of the successive layer and then ultimately be transmitted by the NIC across the cable to the other NIC. Each layer is in fact communicating with its peer on the other system by creating a packet intended for it. This concept includes the physical layer itself which is communicating with the NIC on the other system. The horizontal arrows pertain to the peer-to-peer concept within the stack. They indicate that the DataLink layer of the PC on the left creates a packet for the DataLink layer driver of the PC on the right. It passes the data payload up to the Network layer. This packet was also created by the Layer 3 driver of the PC on the left explicitly for the layer 3 driver of the PC on the right. The Layer 4 driver of the PC on the left communicates directly with the Layer 4 driver on the right and so forth until the operating system of the PC on the left can interact directly with the operating system of the PC on the right, such that the PC on the left can copy a file from the PC on the right "seamlessly" without worrying about the details of how this actually gets accomplished by the coordinated cooperation of the two network stack drivers communicating via their protocols with each other.

  5. The data link layer consists of two major sublayers: The MAC - Media Access Control sublayer which is responsible for low level addressing of each transceiver (node) and is also responsible for obeying the "rules of contention" or the behavior of the Ethernet transceiver that is sharing the network with an unknown other number of hosts. This includes, for example, determining in coordination with the physical circuitry of the card if the cable is not currently carrying another transmission and is therefore free to initiate a transmission. This layer of the driver program will also respond to detected collisions calculating the backoff period before retry and sensing of the traffic-free line before making this attempt.

  6. The LLC - Logical Link Control layer was first formally defined in the IEEE 802.2 specification and was incorporated into the larger specification for all of Ethernet published as the IEEE 802.3 document. The LLC layer is responsible for the construction of Ethernet frames suitable for transmission when passed a block of data through its "well known" service access point from the network layer above. The LLC layer performs the reverse activity, decoding an arriving frame and passing the data within it up to the network layer above through its well known SAP as well. Frames intended for other nodes are discarded in this layer. Damaged frames are detected here and can be asked for again by this layer. These are just two more functions of the LLC layer.

  7. An Ethernet frame has a specific layout. This layout facilitates the ability of the card to detect it at all, and be able to synchronize its circuitry to the transmission frequency which is crucial to its ability to detect the stream of ones and zero's embedded within it using a frequency dependent data encoding method know as Manchester Encoding. The frame's layout is also specifically designed to be lightweight (fast) and finally to be able to receive the information that it contains accurately and correctly. Furthermore the frame contains information that is relevant to the DataLink layer protocol driver, whose task it is to interpret the frame and determine: 1) Is the frame intended for this system, or some other system on the collision domain, 2) If it is intended for this system, what protocols are being used within the frame, 3) If this driver understands the protocol being employed, pass the data block contained in the frame up to the next layer.

  8. Here is the structure of an Ethernet frame:
    Field:  Preamble   SFD   Dest   Source   Length   LLC/type   Data   Pad   FCS 
    Size: 7 bytes 1 6 6 2 * 46-1500 * 4
    Preamble – A series of alternating 1's and 0's which allows the card's circuitry to perfectly tune to the frequency of the transmission and receive the Manchester encoded data correctly.
    SFD - Start Frame Delimiter; ends the 56 bit preamble with the byte 10101011, indicating that the preamble has ended and the subsequent fields are to be expected.
    Dest – The six byte address of the destination node. This is the Media Access Control (MAC) address which is normally embedded into the card's ROM memory.
    Source – The six byte address of the source node. This is the MAC address of the system that has transmitted this packet.
    Length – The 2 byte value that holds the total length of the data within the frame. (The following field is included within this field.)
    LLC/type – This information is included within the two bytes allocated to the preceding length field. Since the maximum length of the data in the frame is 1500, this is a number that can be expressed as an 11 bit number (211 = 2048) the other 5 bits could be used to encode further frame type information. Such as whether the frame carries an IP packet or an IPX packet. Or if it is an "802.2 Raw" frame type or an Ethernet SNAP, etc.
    Data - The actual data that is being sent within the packet follows the Length/LLC-type bytes and can be anywhere from 46 to 1500 bytes of information.
    Pad – In the event that the data is undersized such that the entire packet falls under a total size of 64 bytes, the pad field would be used to ensure that the entire frame size is forced up to 64 bytes so that the frame will appear to be a valid transmission to the nodes on the collision domain.
    FCS – The Frame Check Sequence (or Sum?) is a 4 byte Cyclical Redundancy Check number that can be recalculated by the card to be certain that the frame has arrived in tact.

  9. The recognized LLC types that can be placed in the LLC-type field of a standard Ethernet frame are Novell IPX or IP. This in essence serves as a rudimentary service access point. In subsequent years SAPs have evolved considerably and these in particular are not that necessary but they will still be honored by the drivers. The preamble and start frame delimiter are not considered vital parts of the frame and are immediately discarded (since they are always the same and it is understood that the preamble is a "Can I have your attention please!" and not all cards will respond fast enough to record every single bit in it. So the minimum 64 byte frame starts with the destination Mac address and ends with the last byte of the FCS. The FCS is calculated against the information stretching from the destination MAC address to the last data byte and of course cannot include itself.

Review Questions

  1. Describe the file LSL.COM:

  2. Describe the file 3C90X.COM:

  3. Describe the file IPXODI.COM:

  4. Describe the file NETX.COM:

  5. Explain why in Windows 9x the line lastdrive=H must be put into the CONFIG.SYS:

  6. Explain the difference between being logged in and attached to a Novell server:

  7. Explain the difference between NDIR and DIR:

  8. Describe two differences between NCOPY and COPY:

  9. Why must NETX be unloaded from memory prior to launching Windows and how is this done?

  10. What four different networking layer components can be added within the Network Neighborhood Properties sheet of Windows 98:

  11. List the default networking components that Windows 98 will set up when a NIC's driver's are installed:

  12. List the necessary operations to change from these defaults to being able to attach and login to the Novell 3.12 server used in the classroom:

  13. Calculate the minimum number of frames that the server will need to transmit in order to send the Win98 image located to your system:

  14. List and describe the fields of an Ethernet frame:

  15. List which field(s) are not included in the official size calculation of the frame and explain why:

  16. List which field(s) are not included in the FCS calculation and explain why:

  17. Which two LLC-types have been formally defined:

  18. Explain the steps to map a network drive and explain what this means:

  19. Explain the decision to set the frame type while configuring IPX/SPX. Explain why the system won't find the server if this is set wrong:

  20. Name the two sublayers of the data link layer and explain what each layer does:

  21. List two specific things that the LLC layer can do with a frame that has arrived:

  22. On a separate sheet list the Novell Netware workstation DOS executable commands that you have learned including any switches, describe what they do and record their screen output.

Copyright©2000-2007 Brian Robinson ALL RIGHTS RESERVED