Introduction to Metropolitan Area Networks and Wide Area Networks

Transcription

1 Introduction to Metropolitan Area Networks and Wide Area Networks Chapter 9 Learning Objectives After reading this chapter, you should be able to: Distinguish local area networks, metropolitan area networks, and wide area networks from each other Identify the characteristics of metropolitan area networks and explain how they compare and contrast with wide area and local area networks Describe how circuit-switched, datagram packet-switched, and virtual circuit packetswitched networks work Identify the differences between a connectionless network and a connection-oriented network and give an example of each Describe the differences between centralized routing and distributed routing, citing the advantages and disadvantages of each Describe the differences between static routing and adaptive routing, citing the advantages and disadvantages of each Document the main characteristics of flooding and use hop count and hop limit in a simple example Discuss the basic concepts of network congestion including quality of service Chapter Outline 1. Introduction 2. Metropolitan Area Network Basics a. SONET vs. Ethernet 3. Wide Area Network Basics a. Types of sub-networks Circuit-switched Packet-switched Broadcast b. Connection-oriented versus connectionless network applications

2 4. Routing a. Dijkstra s least cost algorithm b. Flooding c. Centralized routing vs. distributed routing d. Adaptive routing vs. static routing f. Routing examples 5. Network Congestion a. The problems with network congestion b. Possible solutions to network congestion 6. WANs In Action: Making Internet Connections a. A home-to-internet connection b. A work-to-internet connection 7. Summary Lecture Notes Introduction A network that expands beyond a metropolitan area is a wide area network. Wide area networks share a few characteristics with local area networks: they interconnect computers, they use some form of media for the interconnection, and they support network applications. More importantly, however, wide area networks differ from local area networks in a number of ways. Wide area networks include both data networks, such as the Internet, and voice networks, such as telephone systems, whereas local area networks in almost all cases include only data networks. Wide area networks can interconnect thousands, tens of thousands, or more workstations so that any one workstation can transfer data to any other workstation. As the name implies, wide area networks can cover large geographic distances, including the entire earth. Metropolitan Area Network Basics Many of the same technologies and communications protocols found in local area networks and wide area networks are used to create metropolitan area networks. MANs can be used to support high-speed disaster recovery systems and real-time transaction backup systems. They can also provide interconnections between corporate data centers and Internet service providers. Almost all MANs are based on one of two basic forms of supporting technology: SONET or Ethernet. Wide Area Network Basics A wide area network is an interconnection of computers and computer-related equipment that performs a given function or functions and typically uses local and long-distance telecommunication systems. The types of computers used within a wide area network range from

3 microcomputers to mainframes. The telecommunication lines can be as simple as a standard telephone line or as advanced as a satellite system. Typical functions of wide area networks include the bulk transfer of data between two endpoints, electronic mail services, access to database systems, and access to the Internet. Wide area networks can assist with specialized operations in many fields, such as manufacturing, medicine, navigation, education, entertainment, and telecommunications. All wide area networks are a collection of at least two basic types of equipment: a station and a node. Underlying the surface of a wide area network is the communications network. A wide area network may be categorized by the way it transfers information from one end of the communications network to the other. There are three basic types of networks: circuit-switched, packet-switched, and broadcast. The application that runs over a network can be connection-oriented and connectionless. Routing A wide area network s underlying communications network consists of multiple nodes, each with possible multiple connections to other nodes within the network. Each node is a router that accepts an input packet, examines the destination address of the packet, and forwards the packet onto a particular communications line. With multiple-linked nodes there may be one or more paths entering a node as well as one or more paths leaving a node. If most nodes in the communications network have multiple inputs and outputs, numerous routes from a source node to a destination node may exist. To determine a particular route, computer networks perform routing. To get a feel for routing in wide area networks, we examine several of the most commonly used routing techniques: Dijkstra s least cost algorithm, flooding, distributed routing, centralized routing, and adaptive routing. Most wide area networks use a combination of these routing techniques to achieve a routing algorithm that is fair, efficient, and robust, yet stable. Two examples of real Internet routing algorithms include RIP and OSPF. Network Congestion When a network or a part of a network becomes so saturated with data packets that packet transfer is noticeably impeded, network congestion occurs. Congestion may be a short-term problem, such as a temporary line or node failure, or it may be a long-term problem, such as inadequate planning for future traffic needs or poorly created routing tables. Computer networks are going to experience congestion, and no amount of planning can avoid this situation. Thus it is important to consider effective congestion avoidance and congestion handling techniques. WANs In Action: Making Internet Connections The two In Action examples in this chapter demonstrate the types of network connections that are necessary to interface a computer to the Internet both at home and at work.

4 Quick Quiz 1. What are the main differences between circuit-switched networks and packet-switched networks? Circuit-switched: dedicated, physical circuit; all traffic follows this path; charged for duration of circuit Packet switched: not a dedicated, physical circuit; path may be shared by other users; data sent as packets; packets may follow different paths 2. What are the main differences between connectionless networks and connection-oriented networks? Connectionless: no connection is made before data is transmitted; very simple error control, if any Connection-oriented: connection must be made before any data sent; error control often more elaborate 3. What are the basic routing techniques? Dynamic, centralized, distributed, least cost, flooding Discussion Topics 1. Why should systems dealing with money be connection-oriented? 2. Are there any examples of everyday communication systems that use flooding? Teaching Tips 1. Be sure to emphasize how the communications sub-network describes the underlying technology of a network and is either packet-switched, circuit-switched, or broadcast, and the application that runs over the network is either connectionless or connection-oriented. Solutions to Review Questions 1. What are the main differences between a local area network and a wide area network? Local area network: smaller, different protocols, different functions

5 2. How does a metropolitan area network differ from a wide area network? How are they similar? Metropolitan area networks cover a much smaller area than wide area networks. Both use similar technologies and media. 3. What is meant by failover time? The time it takes for a network to reroute itself to a known good path 4. What are the two types of technology that support metropolitan area networks? Ring-based SONET and Ethernet 5. What is Metro Ethernet, and how does it relate to metropolitan area networks? Metro Ethernet is a service that connects businesses to other businesses using an Ethernet structure. 6. What is a sub-network and how does it differ from a wide area network? The sub-network is the underlying structure of nodes and communication links. The network includes users, end stations, and applications. 7. What is the difference between a station and a node? A station is where a user interfaces to the network. A node is a connection point within a network or the connection point between a station and the network. 8. What are the main characteristics of a circuit-switched network? What are its advantages and disadvantages? Characteristics: dedicated, fixed circuit Advantages: path is dedicated to only you; no routing decisions necessary Disadvantages: can be costly, especially if you are not transmitting continuously 9. What are the main characteristics of a datagram packet switched network? What are its advantages and disadvantages? Each packet is sent on its own; each packet requires its own routing decision and may take a different path from other packets. Advantages: fast rerouting, dynamic; Disadvantage: each packet needs its own routing decision. 10. What are the main characteristics of a virtual circuit packet-switched network? What are its advantages and disadvantages?

6 Characteristics: Data are sent in packets; all packets follow same path created via tables; path may be shared with other users. Advantages: more efficient use of paths; no routing decisions necessary (except the first one that creates the virtual circuit) Disadvantages: if a path experiences problems, network cannot easily reroute 11. How does a connectionless network application differ from a connection-oriented network application? Connectionless is simpler, quicker, but potentially less reliable. 12. Is a connectionless network application reliable or unreliable? Explain. Unreliable. A connection is not established before transmission begins, and individual packets are not acknowledged. 13. What are the various combinations of circuit-switched and packet-switched networks and connection-oriented and connectionless network applications? Circuit-switched connection-oriented Circuit-switched connectionless Packet-switched connection-oriented Packet-switched connectionless 14. How does a weighted network graph differ from a network graph? Each edge or path between nodes has a weight or cost associated with it. 15. For a weighted network graph, how many different definitions of weight can you list? Transmission speed (delay), buffer sizes, link cost, other equipment cost 16. What are the basic goals of Dijkstra s least cost algorithm? To find the least cost path amongst all paths from a starting node to all other nodes in the network 17. How can flooding be used to transmit a data packet from one end of the network to another? Determine how many hops it should take to get a packet from one end to the other and set the hop limit to that value. 18. How are the hop count and the hop limit used to control flooding?

7 Every time a packet hops to the next node, its hop count is incremented. When a packet s hop count equals the hop limit, the packet is discarded. 19. What are the main advantages and disadvantages of: a. Centralized routing b. Distributed routing c. Adaptive routing a. All tables are kept in one location; what happens if one location fails or is a bottleneck? b. No single location is prone to failure; many messages are needed to perform distributed operation. c. Can adapt to changes in network but time-consuming and tricky to update tables 20. What are the differences between RIP and OSPF? RIP is simpler but does not adapt quickly to network changes; OSPF is more complex but adapts more quickly. 21. What can cause network congestion? Too much traffic, network failures, or problems 22. How can network congestion be avoided? Limit the amount of traffic; don t let a new application start unless you can support it. 23. What does quality of service have to do with network congestion? Users request a particular level of quality of service. Network should be sure it can support user request before accepting; if network cannot support request, congestion may occur. Suggested Solutions to Exercises 1. State three advantages of a SONET-based metropolitan area network over an Ethernetbased metropolitan area network. SONET is well-established, very common, high-speed, with small failover times. 2. State three advantages of an Ethernet-based metropolitan area network over a SONETbased metropolitan area network. Ethernet is simpler, less-expensive, and supports bursty traffic better. 3. Which type of network application requires more elaborate software: connectionoriented or connectionless? Explain.

8 Connection-oriented; must create a dialog that establishes, maintains, and deletes the connection 4. Create an analogy similar to the telephone call/sending-a-letter scenarios that demonstrates the differences between connection-oriented and connectionless network applications. Connectionless: sending an Connection-oriented: taking a package to FedEx 5. Explain the difference between a network node and a network station. A node is the routing point within a communications network or the access point to the communications network. A station is the point at which the user interfaces. 6. Does a datagram sub-network require any setup time before a packet is transmitted? If so, when and how often? No. Each packet is routed on its own. 7. Does a virtual circuit sub-network require any setup time before a packet is transmitted? If so, when and how often? Yes. The virtual route has to be created by passing information to each node along the route. 8. List the steps involved in creating, using, and terminating a virtual circuit. Creating: Virtual route is determined and each node along the route is informed. Nodes routing tables are updated accordingly. Using: As a packet arrives, if it belongs to a particular virtual route, its routing information is extracted from the routing tables. Terminating: Virtual route routing tables can be discarded. 9. You are downloading a file over the Internet. Is the download a connectionless application or a connection-oriented application? More than likely, the download is a connection-oriented operation. 10. Using flooding and the graph shown in Figure 9-17, how many packets will be created if a packet originates at node A and there is a network hop limit of three? 30 packets (three copy 1s, seven copy 2s, 20 copy 3s) 11. How do you determine the hop limit in flooding? Choose the two points farthest apart and determine how many hops it should take for a packet to traverse between the two points.

9 12. Explain how RIP may gather the wrong or insufficient data. Essentially RIP is good at transferring positive data router by router, but not so good at transferring negative data. The algorithm always looks for a better path. If a neighboring router presents negative data (a path getting worse), it may be ignored. 13. What can be done to protect a centralized routing network if the computer hosting the central routing table fails? Select another node as a backup site 14. What happens in a virtual circuit packet-switched network if a node or communications link along the virtual path fails? A new virtual circuit has to be created. 15. Does an Ethernet network support quality of service? Explain. Basic Ethernet does not. It is a first-come, first-served algorithm. Wireless Ethernet can favor certain devices by using the interframe spacing concept. 16. Frame relay supports two kinds of congestion avoidance procedures. Does the destination station learn of the congestion in both procedures? Only in forward explicit congestion notification is the destination station informed. 17. Assume that two stations on a typical ATM network that is transmitting data at a speed of 150 Mbps are at opposite ends of the United States (the propagation time to transmit between the two stations is approximately 46 x 10-3 seconds). Congestion begins to occur at one station. If the congested station wants to send a signal back to the other station to slow down or stop, how many bits will be dumped before the other station slows down or stops its transmission? If the one station sends a slow-down message to other station, it will take 46 x 10-3 seconds for the other station to receive the message. During that time, the sending station will send 150,000,000 bits per second * 46 x 10-3 bits, or 6,900,000 bits.

10 Thinking Outside the Box 1. Your company has a number of interconnected local area networks within a single building. Within the building is a router that connects the LANs to an Internet service provider, using a couple of T-1 telephone lines. Your company is considering replacing the T-1s with a connection to a metropolitan area network. Show the interconnection of the LANs, the MAN, and the Internet, assuming the MAN is SONET-based. Would the interconnection be different if the MAN were Ethernet-based? Explain. As far as your drawing is concerned, they are essentially the same. T-1s would require CSU/DSU, while an Ethernet-based connection might connect directly to a high-speed campus switch. 2. A large wide area network covers the United States and has multiple nodes in every state. The node in Denver crashes completely. How do the other nodes in the country find out about the crash if RIP is used? If OSPF is used? Information eventually filters back to all nodes in an RIP network. Information should move more quickly in OSPF. 3. Your company is creating a new network application that allows employees to view their pay stubs electronically via the Internet. Should this application be connectionless or connection-oriented? Defend your answer, and draw a series of sample screens that a user accessing his or her electronic pay stub might view. You would want a connection-oriented application for something as important as viewing electronic pay stubs. 4. One form of congestion avoidance is the permit system, in which a node has to have a permit before it can transmit. Suppose a wide area network is using the permit system to control congestion. What happens if, for some unknown reason, all the permits disappear? How can this event be detected? How can this event be repaired? Management of permits is tricky. How would you keep track of where the permits are? How would you redistribute if you had to?