Notes On Chapter Thirteen -- Local Area Networks: Packets, Frames and Topologies

• 13.0 Study Guide
  
  
  • Understand the concept of packet switching, what are its advantages, and how it differs from circuit switching.
  • Understand what the data link layer is, and where it is located in the TCP/IP protocol stack
  • Have some familiarity with the IEEE family of 802.x standards for LANs.
  • Be familiar with the major LAN topologies
  • Know the basics of the IEEE 48-bit MAC addressing scheme
  • Understand what unicast, broadcast, and multicast addresses are
  • Know the typical structure of a frame in a packet-switched LAN
  
  
• 13.1 Introduction
  
  
  • IEEE Standards Model
  • Hardware Addressing
  • Frame Identification
  
  
• 13.2 Circuit Switching
  
  
  • Circuit switching is: "a communication mechanism that establishes a path between a sender and a receiver with guaranteed isolation from paths used by other pairs of senders and receivers.
  • Nowadays circuits are virtual - multiplexed over shared media
  • The properties that define a circuit switched paradigm
    • Point-to-point communication
    • Separate steps for circuit creation, use and termination
    • Performance equivalent to an isolated physical path (virtual circuits use FDM or synchronous TDM. )
  
  
  
  
• 13.3 Packet Switching
  
  
  • Packet switching uses statistical multiplexing.
  • Senders must divide their messages into small blocks of data (packets).
  • Properties that define a packet-switched paradigm
    • Arbitrary, asynchronous communication (implies many-to-many communication)
    • No set up required before communication begins
    • Performance varies due to statistical multiplexing among packets
  • Packet switching allows senders to send to multiple recipients concurrently, and allows receivers to receive from multiple senders concurrently.
  
  
  
  
• 13.4 Local and Wide Area Packet Networks
  
  
  • LAN - Local Area Network - least expensive - spans single room or building
  • MAN - Metropolitan Area Network - Medium expense - spans major city or metroplex
  • WAN - Wide Area Network - Most expensive - spans sites in multiple cities
  
  
  
  
• 13.5 Standards For Packet Format and Identification
  
  
  • Packets have to contain the address of the intended recipient.
  • Protocols have to stipulate the form of address and placement of the address in the packets.
  • The Institute for Electrical and Electronic Engineers (IEEE) created the most widely used set of LAN standards.(Project 802 LAN/MAN Standards Committee, 1980)
  
  
  
  
• 13.6 IEEE 802 Model and Standards
  
  
  • IEEE divides the Data Link layer into two sub-layers.
  • The Logical Link Control sub-layer specifies addressing and the use of addresses for demultiplexing.
  • The Media Access Control sub-layer specifies how a group of computers share the medium.
  • The LAN protocols are numbered starting with the suffix 802.
  
  
  
  
  
  
• 13.7 Point-to-Point and Multi-Access Networks
  
  
  • In a multi-access network, many computers share a medium in a way that allows any pair to communicate.
  
  
• 13.8 LAN Topologies
  
  
  • 13.8.1 Bus Topology
    • All computers attach to a single cable or other medium
    • When a computer sends, all can receive.
    • A protocol is needed to handle collisions
    • (may resemble a star topology)
    
    
  • 13.8.2 Ring Topology
    • Computers connected in a "chain", plus the last connected back to the first.
    • (may resemble a star topology)
    
    
  • 13.8.3 Mesh Topology
    • A direct connection between each pair of computers
    • High performance is possible.
    • This is very expensive except for very small networks.
    
    
  • 13.8.4 Star Topology
    • All computers attach at one "central" point.
    
    
  • 13.8.5 The Reason For Multiple Topologies
    • Each topology has advantages and disadvantages
    • In a ring it is easy to coordinate access and tell whether a link is down. If one link is down the entire network is disabled.
    • A link lost in a star cuts off only one host from the network.
    • A bus network is inexpensive but everything depends on the one bus working.
  
  
  
  
• 13.9 Packet Identification, Demultiplexing, MAC Addresses
  
  
  • IEEE created an addressing standard for LANs
  • Under the standard, each NIC is assigned a unique 48 bit address
  • Each packet sent on the LAN is required to contain the address of the intended recipient of the packet.
  • These 48-bit IEEE addresses are called Media Access Control (MAC) addresses. They are also called Ethernet addresses, because they originated with the development of Ethernet technology.
  • IEEE regulates the allocation of Ethernet addresses to make sure that NIC's manufactured by vendors are assigned unique addresses.
  
  
  
  
• 13.10 Unicast, Broadcast and Multicast Addresses
  
  
  • There are three types of IEEE MAC (Ethernet) addresses:
    • Unicast - for sending a packet to exactly one recipient
    • Broadcast - for sending a packet to be delivered to every NIC on the network. The broadcast address is "all 1's"
    • Multicast - for sending a packet to each of a specific set of NIC's
  
  
  
  
• 13.11 Broadcast, Multicast and Efficient Multi-Point Delivery
  
  
  • Broadcast and Multicast are relatively efficient on a shared-medium LAN such as Ethernet, because a sender puts only a single copy of the packet on the LAN.
  • All packets placed on the LAN are monitored by every NIC connected to the LAN.
  • The NIC examines the address of each packet
  • If the address is "a match" the NIC "accepts" the packet.
  • The address matches if:
    • it is the (unique) MAC address of the NIC, or if
    • it is the broadcast address, or if
    • it is the multicast address of a multicast group to which the NIC belongs.
  
  
• 13.12 Frames and Framing
  
  
  • Typically a header is added to a packet and sent with it on the network
  • Headers contain things like the address of the intended recipient, a check sum, and other metadata.
  • Typically the packet header is sent first, followed immediately by the packet 'payload' - actual message data.
  • Sometimes a short 'prelude' (prolog) is sent just before the header, and sometimes a short 'postlude' (epilog) after the payload.
  • The prelude and postlude can be helpful when recovering from errors or synchronizing.
  
  
  
  
• 13.13 Byte and Bit Stuffing
  
  
  • Suppose that ASCII SOH and EOT are used as the prelude and postlude.
  • Then those characters can't be allowed inside the packet
  • If they do occur inside the packet, the sender can do byte stuffing.
  • An example is if, before sending, the the sender replaces SOH inside the packet with ESC-A, EOT with ESC-B, and ESC with ESC-C.
  • When the receiver encounters any of {ESC-A, ESC-B, ESC-C} in the received packet, it replaces them with SOH, EOT and ESC respectively.
  • If the sender desires something like ESC B to occur in the data, there is no problem. ESC B is encoded as ESC C B, and the decoding scheme works on that sequence properly.