Networking Technologies

Chapter 7: The OSI Data-Link Layer

 

 

Objectives:

Chapter 7 is the second chapter about OSI model layers. The objectives important to this chapter are on page 7-1:

  1. Understanding the overall purpose of this layer.
  2. Identify the characteristics of the two logical topologies.
  3. Identify the characteristics of the three media access methods.
  4. Identify the nature and uses of the addresses defined and managed at the Data Link layer.
  5. Identify the characteristics of the three transmission synchronization techniques that can be used at the Data Link layer.
  6. Identify the characteristics of the connection services that can be implemented at the Data Link layer.
Concepts:

The picture on page 7-2 tells us part of what this layer is about: how to package bits into frames. This layer tells us how the network is physically accessed, how some errors are handled, how data flow is handled and how entities on the network are addressed.

The topic chart on page 7-3 shows the sub-layers, topics and methods for this chapter.

  • Data-Link - This layer has 2 sub-layers and 5 topics, and is associated with Frames
    • MAC sublayer
      • Logical Topology - 2 Methods
        • Bus
        • Ring
      • Media Access - 3 Methods
        • Contention
        • Token Passing
        • Polling
      • Addressing - 1 Method
        • Physical Device Address
    • LLC sublayer
      • Transmission Synchronization - 3 Methods
        • Synchronous
        • Asynchronous
        • Isochronous
      • Connection Services - 3 Methods
        • Unacknowledged Connectionless
        • Connection Oriented
        • Acknowledged Connectionless

Also note, on page 7-3, that this layer is associated with certain hardware:

  • Bridges
  • Switches (sometimes called Intelligent Hubs)
  • Network Interface Boards (Cards, Adapters, etc.)

In general, the Media Access Control topics are about sharing and connecting to a medium. The Logical Link Control topics are about setting up and maintaining the connections for transmitting frames.

Page 7-5 explains that regardless of the actual Physical topology used, networks also have a Logical topology: the network will either act like a bus or a ring. In a Logical Bus, all devices on a network receive every signal. Then they decide if it is for them or not. In a Logical Ring, devices pass the signal along until it is received by the addressee. Every device on the net does not have to see all signals in a Logical Ring.

Media Access is accomplished in three general ways:

  • Contention - all devices try to send whenever they wish
  • Token-passing - devices send when it is their turn
  • Polling - devices are asked if they have anything to send

Four factors are listed on page 7-8 about Media Access methods. It is recommended that an analyst consider each of these factors when picking a method:

  • Consistent or bursty transmissions - do we send often (consistent) or seldom (bursty)?
  • Amount of data transmissions - large or small amounts in each transmission?
  • Time sensitivity of data - how soon does it have to be transmitted?
  • Number of active devices - how many will the net support?

Contention systems work by letting each device try to send a message on the net as needed, contending or competing with all the other devices for the bandwidth. Two examples of protocols that support such systems are CSMA/CD (Carrier Sense, Multiple Access, with Collision Detection) and CSMA/CA (Carrier Sense, Multiple Access, with Collision Avoidance). A collision occurs when two signals collide on the medium, causing signal loss. These are bursty protocols, best supporting intermittent transmissions. Only about 60 to 70 users can be supported on a segment. Time sensitivity is good, as users do not often have to wait for media access.

In a CSMA/CD system, like Ethernet, the collision is detected and the devices each wait a random number of seconds before sending again. This usually results in one device going ahead of the other. In a CSMA/CA system, devices can be assigned time slices or can be required to ask permission to send, avoiding collisions. Apple LocalTalk is noted as an example of this.

Token-Passing involves passing a token, a small data frame, from station to station. When a station has the token, it is that station's turn to access the medium. Examples of this method are token ring, FDDI (fiber optic rings), and token bus. This type of media access is predictable and consistent, allowing large or small transmissions. It is not the best for time sensitive data since waits are built in, but it will support more devices than contention.

The chart on page 7-12 shows the relative performance of contention nets versus token nets. It shows that contention is best when the load is light, that token passing is better with heavier loads, and that both schemes crash under too much load.

Polling is discussed on page 7-13. You should know it exists, and that it is not used by any major network vendor. It is too slow and controlled to give the users the speed they expect from other network methods. It may be used, however, on a small, time sensitive network, such as one used by automated machinery. The terminology used may show up on a quiz: the unit that polls the others can be called the controller, the primary, or the master. Units that are polled can be called secondaries.

Addressing is discussed in several chapters, and it means something different in each chapter. In this chapter, we are talking about the hardware address, the so called MAC address that is burned into most network cards. You should be aware that ARCnet cards require the user to configure an address on each card, using dip switches. (Anyone want to set up a large ARCnet this weekend?) The use of the address is summarized on page 7-16: the hardware address of a device is placed into a frame, and the frame is sent to all devices on a net. The device whose address matches extracts the data from the frame, while other devices ignore it.

Bridges can read these frame addresses, too. They are programmed to reduce unnecessary LAN traffic, remember? So a bridge will decide whether to pass a frame across media segments based on the address. The bridge keeps a table of where devices are located (on which side of the bridge), so it can send traffic across when necessary and drop it when convenient.

Now, the LLC sublayer. The first topic is Transmission Synchronization. This is for frames, not bits like the last chapter. The frame contains a series of bits, and that series allows us to begin checking for accuracy in transmission. This is beneficial, but also causes overhead in transmission.

Asynchronous transmission requires senders and receivers to synchronize their internal clocks based on signals in the transmission. These are not time signals, but message signals such as start and stop bits for each frame. We also include parity bits, to check accuracy of received frames. A parity bit is either a 1 or a 0, like all bits. An example will make this clearer. The sender counts the number of 1s in a frame. Let's assume the frame looks like this:
1110 0001
This frame has an even number of 1s. If we were using odd parity, we would set the parity bit to 1, to make an odd total for the frame. If we were using even parity, we would set the parity bit to 0, to leave the total as an even number. The way you set the parity bit is determined by your odd or even scheme and the actual number of 1s in each frame.

Synchronous transmission schemes provide a timing signal in the message or on a separate channel. This method allows the sender and receiver to use the same clock. A more complex method of error detection is often used, a Cyclic Redundancy Check (CRC), which builds a two dimensional chart of the frames, and transmits parity bits for every row and column in each chart. The larger the chart used, the more complicated the algorithm (calculation method), and the more secure the method of error trapping.

Isochronous transmission is a newer method. It depends on a separate clock device that transmits synchronization time to all devices on the network. In this way all devices are depending on the same clock. The clock signal is a constant that the sender and receiver know about and use. It is independent of the message signals.

The last topic for this chapter is Connection Services. This topic appears in other chapters as well, and each time it means something specific to that chapter (layer).

First, you need to know that there are three categories of connection services:

  • Unacknowledged Connectionless Services - no flow, error, or packet sequence control. No guarantee of delivery.
  • Connection-oriented Services - flow, error, and packet sequence control. Guaranteed delivery.
  • Acknowledged Connectionless Services - uses acknowledgments for flow and error control. No sequence control needed.

There are two functions of Data Link-LLC services:

  • Flow Control - to control the amount of data sent between computers
  • Error Control - to detect errors and request retransmission

Flow Control tells faster senders to slow down for slow receivers. How it is done gets divided up again:

  • Window Flow Control - sets up a buffer (window) which holds a number of frames. The frames are actually numbered for error control.
    • Static - uses one window size (a static number of frames)
    • Dynamic - window size adjusts up or down, uses a choke packet to signal "too much!"
  • Guaranteed Rate Flow Control - the sender and receiver negotiate a rate before data transmission, which is in effect for the whole conversation.

Error Control allows the receiver to use the frame numbers in a window to let the sender know when a frame was not received or when the CRC check does not match. Then the sender resends the missing or bad frame.