Chapter 3. Internetworking

Transcription

1 Computer Networks: A Systems Approach, 5e Larry L. Peterson and Bruce S. Davie Internetworking Copyright 2010, Elsevier Inc. All rights Reserved 1

2 Problems In Chapter 2 we saw how to connect one node to another, or to an existing network. How do we build networks of global scale? How do we interconnect different types of networksto build a large global network? 2

3 Chapter Outline Switching and Bridging Basic Internetworking (IP) Routing 3

4 Chapter Goal Understanding the functions of switches, bridges and routers Discussing Internet Protocol (IP) for interconnecting networks Understanding the concept of routing 4

5 Switching and Forwarding Store-and-Forward Switches Bridges and Extended LANs Cell Switching Segmentation and Reassembly 5

6 Switching and Forwarding Switch A mechanism that allows us to interconnect links to form a large network A multi-input, multi-output device. Transfers packets from an input to one or more outputs 6

7 Switching and Forwarding Adds the star topology to the point-to-point link, bus (Ethernet), and ring(802.5 and FDDI) topologies 7

8 Network Topologies The topology of a network defines how the nodes of a network are connected. Why topology so important? Solve Internetworking Problems: -routing -resource reservation -administration 8

9 Network Topologies 1. The Bus Topology 2. Star Topology 3. Ring Topology 9

10 Bus Topology H1 H2 H3 H 5 All stations are attached to a single cable Used in Ethernets No central host Can operate even if one of the computers malfunctions Easy to wire Less Expensive A bad connection to the cable can bring down the entire network Collisions possible H4 H6 10

11 Star Topology H 1 H 6 H 2 Switch H 5 H 4 H 3 All communication must pass through the central computer The communications in the network will stop if the switch stops functioning 11

12 RingTopology H 1 H 6 H 2 H 3 H 5 H 4 Used in token ring network. The connecting wire, cable, or optical fiber forms a closed loop. Data are passed along the ring from one computer to another. 12

13 Switching and Forwarding A switch has a fixed number of inputs and outputs. Large networks can be built by interconnecting a number of switches A switch s primary job: switching and forwarding: receive incoming packets on one of its links and transmit them on some other link 13

14 Switching and Forwarding How does the switch decide which output port to place each packet on? It examine at the header of the packet Two common approaches: 1. Datagram or Connectionless approach 2. Virtual circuit or Connection-oriented approach 3. A third approach source routingis less common 14

15 Switching and Forwarding Switched Communication Network Packet-Switched Circuit-Switched Datagram Virtual Circuit 15

16 Switching and Forwarding Assumptions: Each host has a globally unique address Each switch can uniquelyidentify the input and output ports We can use numbers, names 16

17 Datagram Model Datagrams Every packet contains the complete destination address i.e. a packet contains enough information to enable any switch to decide how to get it to destination 17

18 Datagram Model An example network: To decide how to forward a packet, a switch consults a forwarding table(sometimes called a routing table) 18

19 Datagram Model Forwarding Table for Switch 2 Destination Port A 3 B 0 C 3 D 3 E 2 F 1 G 0 H 0 19

20 Datagram Model There is no round trip time delay waiting for connection setup: A host can send data as soon as it is ready Transport is unreliable: Source doesn t know if the network can deliver a packet or if the destination host is even up Packets are treated independently: Each packet follow its own route It is possible to route around link and node failures Every packet must carry the full address of the destination the overheadper packet is higher 20

21 Virtual Circuit Model Also called a connection-oriented model Uses the concept of virtual circuit(vc) Two steps: 1. First set up a virtual connection from the source host to the destination host 2. Then send the data 21

22 Virtual Circuit Model Connection Setup: Reserve resources along a selected path All packet follow the same path Typically wait full RTT for connection setup before sending data Connection request contains the full address for destination Data packets contain only a small identifier called Virtual Circuit Identifier (VCI) i.e. Overhead per packet is small If a switch or a link in a connection fails, the connection is broken and a new one needs to be established 22

23 Switching and Forwarding Host A wants to send packets to host B sender reciever 23

24 Switching and Forwarding Connection setup Establish connection state in each of the switches between the source and destination hosts Each switch maintains a VC Table : Map: incoming connections outgoing connections How to uniquely identify a connection? The combination of a Virtual Circuit Identifier (VCI)(in message) and port/interface the uniquely identifies the connection at a switch The connection state for a single connectionconsistsof an entry in the VC table in each switch through which the connection passes 24

25 Switching and Forwarding One entry in the VC table on a single switch contains AnIncoming VCI: uniquely identifies the incoming connection at this switch Carried in the incoming packet header that belong to this connection An incoming interface : Port/interface on which packets for this VC arrive at the switch An outgoing interface : Port/Interface in which packets for this VC leave the switch An Outgoing VCI: A potentially different VCI than incoming VCI Switch replace incoming VCI by this in outgoing packets 25

26 Switching and Forwarding In following example, when a switch gets a packet it does the following: 1. Receives an incoming packet from interface (=2). 2. Examine packet header to find incoming packet VCI (=5) 3. Look up <incominginterface, IncomiVC> (=<2,5>) entry in VC table 4. Find outgoing interface (=1) and outgoing VCI (=11) 5. Change packet header VCI to outgoing VCI(=11) and forward it through outgoing interface (=1) Incoming connection Incoming Interface Incoming VC Outgoing Interface Outgoing VC Sample VC Table Outgoing connection 26

27 Switching and Forwarding Creating a new virtual connection : Create a new VC Table entry in each switch in path sourcedestination We assign a new VCI on each linkthat the connection will traverse Ensure that the chosen VCI on a given link is not currently in use on that link by some existing connection. Can be done by: 1. Network administrator - permanent (PVC) 2. Host - switched (SVC) 27

28 Connection Establishment Example Let s assume that a network administrator wants to manually create a new virtual connection from host A to host B 1. First the administrator identifies a path through the network from A to B 28

29 Connection Establishment Example 2. The administrator then picks a VCI value that is currently unused on each link for the connection Suppose the VCI value 5 is chosen for the link from host A to switch 1 11 is chosen for the link from switch 1 to switch 2 So the switch 1 will have an entry in the VC table Switch 1 VC Table: Incoming Interface Incoming VC Outgoing Interface Outgoing VC Similarly, switches 2 and 3 are configured VCI=5 VCI=11 29

30 Connection Establishment Example Incoming Interface Incoming VC Outgoing Interface Outgoing VC Switch 1 VC Table: Switch 2 VC Table: Switch 3 VC Table: Incoming Interface Incoming VC Outgoing Interface Outgoing VC Incoming Interface Incoming VC Outgoing Interface Outgoing VC VCI=11 VCI=11 VCI=5 VCI=7 VCI=7 VCI=4 30

31 Data Transfer Example Host A: Puts the VCI value 5 in the header of the packet and sends it to switch 1 Switch 1: Receive:Receives any such packet on interface 2, Map: Uses the combination of the interface and the VCI (2,5) in the packet header to find the appropriate VC table entry. Forward:The table entry tells switch to forward the packet out of interface 1 and to put the VCI value 11 in the header Incoming Interface Incoming VC Outgoing Interface Outgoing VC 31

32 Data Transfer Example Switch 2: Receive: Packet will arrive on interface 3 bearing VCI 11 Map: (3,11) (2,7) Forward:forward packet from interface 2 with VC =7 in header Switch 3: Receive: Packet will arrive on interface 0 bearing VCI 7 Map: (0,7) (1,4) Forward:forward packet from interface 1 with VC =4 in header Host B: To host B, this VC=4 means the packet as having come from host A 32

33 Switching and Forwarding Source Routing All the information about network topology that is required to switch a packet across the network is provided by the source host Rotate header: Message header contain a list of output portsalong the selected path At message receipt each switch: read first item as its output port Rotate header by 1 item. Forward message in output port source Rotate header Rotate header destination 33

34 QUESTION What are the disadvantages of header rotating based source routing? 34

35 ANSWER It assumes host A knows enough about the topologyof the network to form a header that has all the right directions in it for every switch in the path. We cannot predict how big the header needs to be, since it must be able to hold one word of information for every switch on the path. 35

36 Switching and Forwarding Other approaches in Source Routing: (a) rotation (b) stripping Each switch rotates the Each switch strips list after it has read its the first element own entry and use it (c) pointer Header carry a pointer to the next port entry 36

37 QUESTION What devices we use to connect two Ethernets? 37

38 ANSWER: Bridges and LAN Switches 1. Put a repeater in between them It might exceed the physical limitation of the Ethernet No more than four repeaters between any pair of hosts No more than a total of 2500 m in length is allowed 2. put a bridge between Ethernets and Forward frames from one Ethernet to the other A collection of LANs connected by one or more bridges is usually said to form an Extended LAN 38

39 Bridges and LAN Switches Bridges and LAN Switches Class of switches that is used to forward packets between shared-media LANs such as Ethernets Known as LAN switches Referred to as Bridges 39

40 Bridges and LAN Switches Simplest Strategy for Bridges Accept LAN frames on their inputs and forward them out to ALL other outputs Used by early bridges Learning Bridges Observe that there is no need to forward all the frames that a bridge receives 40

41 Bridges and LAN Switches When a frame from host A that is addressed to host B arrives on port 1, there is no need for the bridge to forward the frame out over port 2. source destination How does a bridge learn ports the various hosts reside? 41

42 Bridges and LAN Switches Solution: Download a table into the bridge Who does the download? Human Too much work for maintenance Can the bridge learn this information by itself? Bridge receives a frame on port 1 addressed to host A DO NOT forward the frame out on port 2 42

43 Bridges and LAN Switches Can the bridge learn this information by itself? Yes How? Each bridge inspects the source address in each frame it receives Recordthe information at the bridge and build the table When a bridge first boots, this table is empty Entries are added over time A timeoutis associated with each entry The bridge discards the entry after a specified period of time If the bridge receives a frame that is addressed to host not currently in the table Forward the frame out on ALLother ports 43

44 QUESTION What is the difference between Bridge Vs. Router? 44

45 ANSWER Bridge works in OSI model layer 2 whereas routers work in layer 3. Therefore bridge forward packets using MAC address while router forward packets using IP address. Routers isolate each LAN into aseparate subnet,so each network adapter's IP address will have a different third "octet Example: and are in different subnets. With a Bridge, all your computers are in the same networksubnet. 45

46 Loops Strategy works fine if the extended LAN does not have a loopin it Frames potentially loop through the extended LAN forever 46

47 Loops Example: B1,B4, B6 are in a loop. 1. B4 gets a packet from Ethernet J 2. B4 has no routing table entries for packet destination address 3. So B4 broadcast packet to Ethernets H, I and B6 gets it from I 4. B6 has no matching routing table entries 5. So B6 forward packet to Ethernet G and B1 gets it 6. B1 has no matching routing table entries 7. So B1 forward packet to Ethernet G and B4 gets it 8. B4 still has no matching routing table entries 9. So loop repeats from step loop

48 Bridges and LAN Switches How does an extended LAN come to have a loop in it? 1. Network is managed by more than one administrator E.g. it spans multiple departments in an organization No single person knows the entire configuration of the network A bridge that closes a loop might be added without anyone knowing 2. Loops are built into the network(more likely scenario) Provide redundancy in case of failures A network with no loops needs only one link failure to become splitinto two separate partitions. 48

49 Bridges and LAN Switches Whatever the cause, bridges must be able to correctly handle loops. Solution: Distributed Spanning Tree Algorithm 49

50 Spanning Tree Algorithm Extended LAN represented by a graph that possibly has loops/cycles A spanning tree is a sub-graphof this graph that covers all the vertices but contains no cycles Spanning tree keeps all the vertices of the original graph but throws out some of the edges (a) a cyclic graph (b) a corresponding spanning tree. 50

51 Spanning Tree Algorithm A protocolused by a set of bridges to agree upon a spanning tree for a particular extended LAN IEEE specification for LAN bridges is based on this algorithm Each bridge decides the ports overwhich it is and is not willing to forward frames Removing ports from the topology A bridge might not participate in forwarding frames 51

52 Spanning Tree Algorithm Algorithm is dynamic The bridges are always prepared to reconfigure themselves into a new spanning tree if some bridges fail Main idea Each bridge selects the ports over which they will forward the frames How does a bridge select ports? 52

53 Spanning Tree Algorithm Algorithm selects ports as follows: Each bridge has a unique identifier (B1, B2, B3, ) Bridges select aroot bridge, a set of designated bridges and a set of preferred paths (subset of ports) that make up the spanning tree 1. Selecting Root bridge: Elect the bridge with the smallest id as the rootof the spanning tree The root bridge always forwards frames out over all of its ports 2. Selecting preferred paths: Each bridge computes the shortest path to the root The port in shortest path is selected as it s preferred path to the root 53

54 Spanning Tree Algorithm 3. Selecting a set of designated bridges: Finally, all the bridges connected to a given LAN elect a single designated bridge. Each LAN s designated bridgeis the one that is closest to the root Designated bridge is responsible for forwarding frames toward the root bridge If two or more bridges are equally close to the root: Select bridge with the smallest id as designated bridge If a bridge is connected to more than 1 LAN (normal): Participates in the election of a designated bridge for each LAN it is connected to 54

55 Spanning Tree Algorithm -Example B1 is the root bridge B3 and B5 are connected to LAN A, but B5 is the designated bridge Why? B5 is closer to root(b1) than B3 B5 and B7 are connected to LAN B, but B5 is the designated bridge Why? Both B5 and B7 have equal distance from B1. So select one with smaller id (B5) Extended LAN Spanning tree 55

56 Spanning Tree Algorithm Problem!: Bridges in an extended LAN cannot see the topology of the entire network. Then how do bridges compute spanning tree? Bridges have to exchange configuration messages with each other Based on messages each bridge decides: 1. If it is the root 2. If it is a designated bridge 56

57 Spanning Tree Algorithm Configuration messages contain : 1. Sending Bridge ID : The ID for the bridge that is sending the 2. Root Bridge : The ID of bridge sending bridge believesto be the root 3. Distance: The distancefrom the sending bridge to the root bridge, measured in hops Each bridge records the current best configuration messageit has seen on each of its ports How does a bridge decide if a new configuration is better than current configuration recorded? 57

58 Spanning Tree Algorithm Answer: The new configuration is betterthan the currently recorded information if: It identifies a root with a smaller id or It identifies a root with an equal id but with a shorter distance or The root id and distance are equal, but the sending bridge has a smaller id 58

59 Spanning Tree Algorithm STEPS: Initialization: Initially each bridge assumes it is the root Configuration message sending: Therefore it sends a configuration message on each port withself= rootand giving distance-to-root=0 Configuration message receipt: Bridge receives configuration message over a particular port. It checks if the new message is betterthan the current best configuration message recorded for that port If the new message is better than the currently recorded one, The bridge discards the old information and saves the new information It first adds 1 to the distance-to-root field 59

60 Spanning Tree Algorithm Configuration message forwarding: When a bridge receives a configuration message indicating that it is NOT the root bridge i.e. a message receipt from a bridge with smaller id The bridge stops generating configuration messages on its own ONLY forwards configuration messages from other bridges after adding 1 to the distance field When the system stabilizes: ONLY the root bridge is still generating configuration messages. Other bridges are forwarding these messages only over ports for which they are the designated bridge 60

61 Spanning Tree Algorithm Consider the situation when the power had just been restored to the building housing the following network All bridges would start off by claiming to be the root 61

62 Spanning Tree Algorithm Configuration message format: (Y,d,X) X= Sender Bridge Y=Root Bridge d= distance from X to Y Consider the activity at node B3 62

63 Spanning Tree Algorithm B2 sends B3 : (B2, 0, B2) At B3: 2 < 3, so B3 accepts B2 as root. At B3: B3 adds 1 to the distance advertised by B2 B3 sends B5: (B2, 1, B3) Meanwhile, B1 sends B2: (B1,0,B1) B1 sends B5: (B1,0,B1) At B2: 1<2, so B2 accepts B1 as root 3(B1,1,B2) B2 sends B3: (B1, 1, B2) At B5: 1<5, so B5 accepts B1 as root B5 sends B3: (B1, 1, B5) At B3: 1<3, so B3 accepts B1 as root At B3: B3 notes both B2 and B5 are closer to root(b1) than it is At B3: Thus, B3 stops forwarding messages in both interfaces This leaves B3 with both ports not selected 3(B1,1,B5) 3 1(B2, 0, B2) (B2,1,B3) 2Root = B2 4Root = B1 2Root = B1 1(B1, 0, B1) 2Root = B1 1(B1, 0, B1) 63

64 Internetworking Can we build large heterogeneous networks from bridges and LAN switches? We can build reasonably large LANs using bridges and LAN switches However, Such networks have low scalability and cannot handle heterogeneity. Let us explore some ways to go beyond the limitations of bridged networks: build large, highly heterogeneous networks Such networks are called internetworks. 64

65 Internetworking What is internetwork? An arbitrary collection of networks interconnected to provide some sort of host-host to packet delivery service Network of networks The nodes that interconnect the networks are called routers (gateways). A simple internetwork H- Hosts R- Routers 65

66 Internetworking What is IP? IP stands for Internet Protocol Key tool used today to build scalable, heterogeneous internetworks It runs on all the nodes in a collection of networks and defines the infrastructure that allows these nodes and networks to function as a single logical internetwork A simple internetwork showing the protocol layers 66

67 IP Service Model 1. Packet Delivery Model Connectionless model for data delivery Best-effort delivery (unreliable service) packets are lost packets can be delayed for a long time packets are delivered out of order duplicate copies of a packet are delivered If something goes wrong, the network does nothing 2. Global Addressing Scheme Provides a way to identify all hosts in the network 67

68 Packet Format Version (4): currently 4 (IPV4) Hlen (4): number of 32-bit words in header TOS (8): type of service (not widely used) Length (16): number of bytes in this datagram. Ident (16): used by fragmentation Flags/Offset (16): used by fragmentation TTL (8): number of hops this datagram has traveled (default value=64) Protocol (8): demux key (TCP=6, UDP=17) Checksum (16): of the header only DestAddr & SrcAddr (32) header 32 bit boundaries data 68

69 IP Fragmentation and Reassembly Each network has some MTU (Maximum Transmission Unit) Largest IP datagram that can be carried by a network Different networks have different MTUs. Ethernet (1500 bytes), FDDI (4500 bytes) Solution: Fragment IP packets and reassemblethem if the size>mtu MTU Smaller than the largest packet size allowed. This supports heterogeneity 69

70 IP Fragmentation and Reassembly Strategy: Fragmentation occurs in a router when it receives a datagram that it wants to forward over a network which has (MTU < datagram) Reassemblyis done at the receiving host All the fragments carry the same identifier in the Ident field Therefore reassembling host can identify frames that go together IP does not attempt recover from missing fragments i.e. if all fragments does not arrive at receiver, host discard fragments that arrived 70

71 IP Fragmentation and Reassembly -Example H5 sends a datagram to H8 MTU=1500 Bytes MTU=1500 Bytes receiver MTU=532 Bytes sender MTU=1500 Bytes 71

72 IP Fragmentation and Reassembly -Example H5 sends a datagram to H8 No fragmentation Fragmentation No fragmentation Receiver host reassemble fragments into original datagram 1420 byte datagram (Wireless) Network 1 MTU = 1500 bytes (Ethernet) Network 2 MTU = 1500 bytes (Point-to-Point)Network 3 MTU = 532 bytes (Ethernet) Network 4 MTU = 1500 bytes IP datagrams traversing the sequence of physical networks 72

73 IP Fragmentation and Reassembly -Example Unfragmented packet: 20 bytes Fragmented packets: 1400 bytes Fragmentation always happen on 8-byte boundaries. i.e. Offset field counts 8-byte chunks, not bytes. 20 bytes 512 bytes Flag field M bit =1 more fragments to follow M bit =0 no more fragments to follow offset fragment1=0 bytes in original fragment2=1* 512/8= 64 bytes in original fragment1=2* 512/8= 128 bytes in original 73

74 Global Addresses IP service model provides a addressing scheme. Each node is assigned a globally unique address. Why not use Ethernet/MAC addresses for routing? Ethernet addresses are flat. i.e. They provide very few clues to routing protocols. Global address (IP address) Properties: globally unique hierarchical: (contains 2 parts:) network + host Network: Identifies the network to which the host is attached. Host: identifies each host uniquely in the attached network i.e. hosts in the same network have same network part but different host parts in their IPs 74

75 QUESTION How to assign IP addresses for the routers? 75

76 ANSWER Routers are attached to two or more networks. They need to have an address on each network, one for each network interface. It is more precise to think of IP addresses as belonging to interfaces than to hosts. 76

77 What Do IP addresses look like? Sizes of the two parts are not same for all addresses IP addresses are divided into 3 classes: Each defines a different size network and host parts 1. Class A address 2. Class B address 3. Class C address There are also class Daddresses that specify a multicastgroup Class E addresses that are currently unused 32 bit address = billion addresses, half are A type, ¼ is B type, and 1/8 is C type 77

78 IP Address Classes Class A Address Start with bit 0. 7 bits per network part : Only = 126 class A networks possible Values 0 and 127 are reserved 24 bits per host: Each class A network can accommodate up to (about 16 million) hosts 78

79 IP Address Classes Class B Address Start with bits bits per network part : Around 2 14 class B networks possible 16 bits per host: Each class B network can accommodate up to 65,534 hosts 79

80 IP Address Classes Class C Address Start with bits bits per network part : Around 2 21 class C networks possible Only 8 bits per host: Each class C network can accommodate up to = 254 hosts One host identifier, 255, is reservedfor broadcast, and 0 is not a valid host number 80

81 Dotted Decimal Notation IP address Format Use Dot notation: four decimal integers separated by dots. Each integerrepresents the decimal value contained in 1 byte of the address starting at the most significant. IP address examples In binary = octet1 octet2 octet3 octet4 81

82 Question Find the IP address ranges possible (min and max IPs) for each address class: Class A Class B Class C 82

83 Answer Class A: Min IP: Max IP: Class B: Min IP: Max IP:

84 Class C: Min IP: Max IP:

85 Address Ranges IP Class Address Range Class A Class B Class C Just by looking the decimal value of the first octet, you can find IP address class 85

86 Address Ranges Dotted decimal treats each octet (byte) as an unsignedbinary integer the smallest value, 0 occurs when all bits of an octet are zero (0) the largest value, 255 occurs when all bits of an octet are one (1) dotted decimal addresses range through

87 Special IP Addresses IP defines a set of special address forms that are reserved These special addresses are never assigned to hosts E.g. IP address for the network is denoted by having all bits in host portion 0 E.g (class B) To simplify broadcasting (send to all) IP defines a directed broadcast address for each physical network Pearson Education Inc., Upper Saddle River, NJ. All rights reserved. 87

88 Special IP Addresses This host A host in the same network E.g Host Broadcast on the local network Broadcast in a distant network (Host address all 1 s) E.g Loopback E.g Network Anything 88

89 IP Address Classes Addressing schemes give lot of flexibility: Networks can select class appropriate for it. The original idea: The Internet would consist of: a small number of wide area networks (class A networks), a modest number of site-(campus-) sized networks (class B networks), and a large number of LANs (class C networks). However, it turned out not to be flexible enough. Today, IP addresses are normally classless ; 89

90 Private IP Addresses Each IP node requires an IP address that is globally unique to the IP internetwork. In the case of the Internet, each IP node on a network connected to the Internet requires an IP address that is globally unique to the Internet (public IP addresses). Most of the hosts on the organization's intranet do NOT require direct connectivity to Internet hosts. They do require a specific set of Internet services through application layer gateways: proxy, -server : World Wide Web access, i.e. Most organizations only required a small amount of public addresses for those nodes(such as proxies, routers, firewalls, and translators) that were directly connected to the Internet. 90

91 Private IP Addresses The private address space specified in RFC 1918 is defined by the following three address blocks: /8 private network : class A network IDs Range of valid IP addresses: to /12 private network: class B network IDs range of valid IP addresses: to /16 private network: class C network IDs range of valid IP addresses: to

92 Private IP Addresses Private addresses are not reachable on the Internet. Therefore, Internet traffic from a host that has a private address must either : Send its requests to an Application layer gateway (such as a proxy server), which has a valid public address, or Have its private address translated into a valid public address by a network address translator (NAT) before it is sent on the Internet. 92

93 Question To which class the following IP address belong?

94 IP Address Format class C 94

95 Question What is its network ID and host ID of IP address ? 95

96 IP Address Format Byte1=202 Byte2= 3 Byte3=2 Byte4= Network ID= Same for all hosts in the network Host ID = 4 96

97 IP Datagram Forwarding Strategy Every datagram contains destination's IP address If datagram receiver node (host/router) is directly connected to destination network, then : forward to host if datagram receiver node is not directly connected to destination network, then: forward to some router each host has a default router each router maintains a forwarding table forwarding table mapsnetwork number into next hop Sample forwarding table 97

98 Question How does a node find out if it is directly connected to the destination network of datagram? 98

99 Answer Node compares the network part of the destination address in received datagram with the network part of the address of each of its network interfaces. Note: Hosts normally have only one interface, Routers normally have two or more 99

100 IP Datagram Forwarding Algorithm if (NetworkNum of destination = NetworkNum of one of my interfaces) then deliver packet to destination over that interface else if(networknum of destination is in my forwarding table) then deliver packet to NextHop router else deliver packet to default router For a host: this simplifies: only one interface and only a default router in its forwarding table if(networknum of destination = my NetworkNum) then deliver packet to destination directly else deliver packet to default router 100

101 Drawbacks of Original IP Addressing Network part in a IP address uniquely identify one physical network This has several drawbacks: 1. Address assignment inefficiency: Imagine a large campus with lot of internal networks No matter how small, each need at least class C address. Network with 2 nodesneed class C address! Waste 253 perfectly useful addresses Network with 256 nodes need class B address! Waste over 64,000 addresses. Why bother? We have 4 million IP addresses possible: It is still finite. There are fewer class B addresses (only 16,000) than class C. Class B addresses have high demand World is running out of IP addresses! 101

102 Drawbacks of Original IP Addressing 2. Routing State maintenance problem: Forwarding table of routers contain entry per network node. So, more network numbers in use bigger forwarding tables Bigger forwarding tables: Add cost to routers Degrade router performance (slow search) 102

103 Routing Table growth ( )

104 Subnetting Therefore we need to assign network numbers more carefully. Solution: subnetting 104

105 Subnetting An IP address has 2 parts: Network ID, Host ID: Subnetting allow you to borrow bits from the Host portion to allow for more Networks: Network prefix Subnetting does NOTgive you more hosts, but actually costs you hosts. Subnetting reduces the size of the routing tables stored in routers. 105

106 Using Host IDs to Subnet Class B Network Subnet Host ID Portion of host address is used for subnet address Subnet Max address: Min address: Subnet 3 Third octet is now used for subnet IDs Ref: 106

107 Subnetting Rules govern this borrowing, ensuring that some bits are left for a Host ID. 1. Two bits MUST remain available to use for the Host ID 2. All of the subnet bits CANNOT be all 1s or 0s at the same time. For each IP address class, only a certain number of bits can be borrowed from the host portion for use in the subnet mask Address Host Bits Available for Class Bits Subnet Class A Class B Class C

108 Routing of Traffic Allow a single network number to be shared among multiple networks Subnets visible only within site Subnets Outside world Routing Ref: 108

109 Subnetting Example Consider the case of a class Caddress assigned to an organization Assume that three of the higher-order bits of the host ID are to be reserved for subnet ID Ref: 109

110 Network ID Sub Net Last Octet Subnet ID Host IP ranges Subnet zero Not used All-ones Subnet Ref: 110

111 Subnet Masking Subnetting, requires a special process called Subnet Masking. The subnet masking identify & extract the Network part of the address. A subnet mask is a 32 bit binary pattern of 1s & 0s that the router uses to cover up the network address to show which bits are being used to identify the subnet. Every IP address is associated with a subnet mask bits borrowed from original host portion for subnet ID 111

112 Subnet Masking Subnet mask is applied to an IP address to determine whether the address is inside/outside the local Network. It is applied to a message s destination IP addressto extract the destination network address. If the destination network address matches the local network ID: the destination is located on the local network. if they don t match: the message must be routed outside the local network. The process of applying the subnet mask involves Boolean Algebra: Use booleanand (&) Destination Network Address = message_receiver_subnet_mask & message_destination_ip_address 112

113 Subnet Masking When a host wants to send a packet to a given IP address: 1. calculate host_subnet_mask & destination_ip_address 2. If result = host_subnet_number: Destination host is in same subnet as sending host Send packet directly to destination 3. Else: Destination host is in a different subnet than sending host Send to default router to be forwarded a different subnet 113

114 Default Subnet Masks There are default standard subnet masks for Class A, B and C addresses: IP Address Class Subnet Mask Class A Class B Class C E.g. Class A IP Address : Subnet Mask : Network ID : Peter Smith 114

115 Question For IP address : associated with subnet mask , what is the subnet address? 115

116 Answer IP = Mask = && = The subnet ID =

117 Subnetting, Subnet & Subnet Mask Subnetting, a subnet & a subnet mask are all different. In fact, the subnettingcreates the subnets& is identified by the subnet masks. Subnetting: The process ofdividing a network & its IP addresses into segments, each of which is called a subnetworkor subnet. Ref: 117

118 Subnetting Example: When the network administrator divides the network into 5 smaller networks (subnets) , , , & the outside world stills sees the network as , but the internal routers now break the network addressing into the 5 smaller subnetworks. Only a single IP address is used to reference the network & instead of 5 network addresses, only one network reference is included in the routing tables of routers on other networks. i.e Subnets visible only within site E.g. Ref: 118

119 Question For the following class A subnet masks, find out 1. the number of bits in the mask 2. Number of subnets 3. Number of hosts per subnet a) b) c)

120 Answer Subnet mask = (no bits borrowed from host portion) No. of 1 bits in mask = 8 No. of subnets = 0 No. of hosts per subnet = = 16,777,214 Subnet mask = (2 bits borrowed from host portion) No. of 1 bits in mask = 10 No. of subnets = = 2 (00 and 11 not allowed) No. of hosts per subnet = = 4,194,

121 Subnetting Remember that the addresses with all ones (broadcast address) & all zeroes (local network) cannot be used as they have special meanings. Class A subnet Mask Number of 1 bits in mask Number of subnets Number of hosts per subnet (default) ,777, ,194, ,048, , , ,094 4, , ,048, ,192,302 2 Ref: peter.joseph.smith@tafensw.edu.au 121

122 Subnetting Class B and C subnet masks are similar The only differences are that you have fewer options (due to a fewer number of bits available) & that you re much more likely to work with Class B & Class C networks in real life. Class B subnet Mask Number of 1 bits in mask Number of subnets Number of hosts per subnet (default) , , , , ,382 2 Ref: peter.joseph.smith@tafensw.edu.au 122

123 Subnetting Class C subnet Mask Number of 1 bits in mask Number of subnets Number of hosts per subnet (default) Notice that: Number of hosts per subnet = (2 number of bits used for host ) -2 Number of subnets = (2 number of bits used for subnets )

124 IP Address Utilization 124

125 QUESTION What unix network commands can you use to find subnet mask? 125

126 ANSWER ifconfig netstat -rn 126

127 Classless Addressing Classless Inter-Domain Routing (CIDR) RFC1338 Counterpart of subnetting Do not use classes to determine network ID Assign any range of addresses to network Why isn t subnetting alone sufficient? Subnetting only allows to split a classful address among multiple subnets. CIDR allows us to combine several classful addresses into a single supernet. (better solution for address space inefficiency) 127

128 Classless Addressing A technique that addresses two scaling concerns in the Internet The growth of backbone routing table as more and more network numbers need to be stored in them Potential exhaustion of the 32-bit address space Address assignment efficiency Arises because of the IP address structure with class A, B, and C addresses Forces us to hand out network address space in fixed-size chunks of three very different sizes A network with two hosts needs a class C address o Address assignment efficiency = 2/255 = 0.78 A network with 256 hosts needs a class B address o Address assignment efficiency = 256/65535 =

129 Classless Addressing Exhaustion of IP address space centers on exhaustion of the class B network numbers Solution Say NO to any Autonomous System (AS) that requests a class B address unless they can show a need for something close to 64K addresses Instead give them an appropriate number of class C addresses What is the problem with this solution? 129

130 Classless Addressing Problem with this solution: Excessive storage requirement at the routers. If a single AS has, say 16 class C network numbers assigned to it, Every Internet backbone router needs 16 entries in its routing tables for that AS This is true, even if the path to every one of these networks is the same If we had assigned a class B address to the AS The same routing information can be stored in one entry Efficiency = / 65, 536 = 6.2% As: Autonomous system -a network that is administered independently of other ASs 130

131 Classless Addressing CIDR uses aggregate routes to solve this. Uses a single entry in the forwarding table to tell the router how to reach a lot of different networks Breaks the rigid boundaries between address classes 131

132 Classless Addressing Consider an AS with 16 class C network numbers. Instead of handing out 16 addresses at random, hand out a block of contiguous class C addresses Suppose we assign the class C network numbers from through Observe that top 20 bits of all the addresses in this range are the same ( ) We have created a 20-bit network number (which is in between class B network number and class C number) 132

133 Classless Addressing Requires to hand out blocks of class C addressesthat share a common prefix The convention is to place a /X after the prefix where X is the prefix length in bits E.g. the 20-bit prefix for all the networks through is represented as /20 By contrast, if we wanted to represent a single class C network number, which is 24 bits long, we would write it /24 133

134 Classless Addressing How do the routing protocols handle this classless addresses It must understand that the network number may be of any length Represent network number with a single pair <length, value> All routers must understand CIDR addressing 134

135 Classless Addressing Route aggregation with CIDR Eight 24-bit network prefixes. Prefixes all start with the same 21 bits. 135

136 IP Forwarding Revisited IP forwarding mechanism assumesthat it can find the network number in a packet and then look up that number in the forwarding table We need to change this assumption in case of CIDR CIDR means that prefixes may be of any length, from 2 to 32 bits 136

137 IP Forwarding Revisited It is also possible to have prefixes in the forwarding tables that overlap Some addresses may match more than one prefix E.g. We might find both (a 16 bit prefix) and (a 24 bit prefix)in the forwarding table of a single router A packet destined to clearly matches both prefixes. The rule is based on the principle of longest match in this case A packet destined to would match and not

138 IP Forwarding Revisited ISP's block /20 Organization /23 Organization /23 Organization / Organization /23 138

139 Address Translation Protocol (ARP) IP datagrams have IP addresses. Datagram receivers hardware interfaces only understand physical (MAC) addresses ARP: Map IP addresses into physical addresses 139

140 Address Translation Protocol (ARP) Techniques 1. encode physical address in host part of IP address E.g. node physical address = = in decimal So give host IP = Problem! : node s physical address should be less than16bits class C only allow 8 bits per host! Clearly doesn t work for a 48-bit Ethernet address. 2. table-based Better solution Each node maintains a table of <IP, physical> address pairs. Table maps IP address into physical address Nodes dynamically build table This is exactly what ARP does! 140

141 Address Translation Protocol (ARP) ARP (Address Resolution Protocol) table of IP to physical address bindings If a host want to send a datagram to a host/router in the same network it checks for mapping in the cache. broadcastrequestif IP address not in table target machine responds with its physical address table entries are discarded periodically (every 15 minutes) if not refreshed (reset timeout) Why? Mappings may change over time E.g. Ethernet card in a host break down and is replaced by a new one with a new address Set of mapping in ARP table is called ARP cache. 141

142 Address Translation Protocol (ARP) Broadcast request for a target IP also include the IP address and the link-layer/physical address of sending host. Target will update its table (add or refresh existing mapping) based on broadcast message information Why? Sending host may be about to send a application level message Target may have to send an ACK to sending host 142

143 ARP Packet Format HardwareType: type of physical network (e.g., Ethernet) ProtocolType: type of higher layer protocol (e.g., IP) HLEN & PLEN: length of physical and protocol addresses Operation: request or response Source/Target Physical/Protocol addresses 143

144 Address Translation Protocol (ARP) Linux commands to check ARP cache arp a -n 144

145 ARP conversation captured with tcpdump 145

146 Host Configurations Ethernet addresses: configured into network by manufacturer They are unique IP addresses: must be unique on a given internetwork must reflect the structure of the internetwork Most host Operating Systems provide a way to manually configure the IP information for the host Drawbacks of manual configuration A lot of work to configure all the hosts in a large network (walk from host to host with a list of available addresses in hand) Configuration process is error-prone Need to ensure no hosts get same IP, each host in network has same network number Automated Configuration Process is required DHCP- Dynamic Host Configuration Protocol does exactly that 146

147 Dynamic Host Configuration Protocol (DHCP) DHCP server is responsible for providing configuration information to hosts There is at least one DHCP server for an administrative domain DHCP server maintains a pool of available addresses An available address is automatically retrieved when a node joins the network Administrator still pick the addresses. He just store them in the server Configuration information of each host is stored in a table indexedby a client ID (usually MAC address) More sophisticated DHCPs allow server to hand out IP addresses on demand. Administrator now only has to allocate a range of IP addresses for each network number. 147

148 DHCP Newly booted or attached host sends DHCPDISCOVER messageto a special IP address ( ) ( ) -broadcastin local network DHCP server has at least one relay agent in per network DHCP relay agent unicaststhe message to DHCP server and waits for the response DHCP messages are sent over UDP 148

149 Internet Control Message Protocol (ICMP) RFC 792 Defines a collection of error messagesthat are sent back to the sourcehost whenever a router or host is unable to process an IP datagram successfully Destination host unreachable due to link /node failure Reassembly process failed TTL had reached 0 (so datagrams don't cycle forever) IP header checksum failed ICMP-Redirect One of most important control messages defined From router to a source host With a better route information E.g. a host has access to two routers(r1.,r2) and send datagram to R1(default router). If R1 knows R2 is a better path for destination it send ICMP-Redirect to host asking to use R2 in future datagrams for destination 149

150 ICMP ICMP provides the basis for two widely used debugging tools: pingand traceroute pinguses ICMP echo messages to determine if a node is reachable and alive. tracerouteuses a slightly non-intuitive technique to determine the set of routers along the path to a destination. 150

151 ICMP Try: ping local host ping traceroute Change max hops and see 151

152 Traceroute example 152

153 Routing Forwarding versus Routing Forwarding: to select an output port based on destination address and routing table Routing: process by which routing table is built 153

154 Routing Forwarding table VS Routing table Forwarding table Used when a packet is being forwarded and so must contain enough information to accomplish the forwarding function A row in the forwarding table contains the mapping from a network number to an outgoing interface and some MAC information, such as Ethernet Address of the next hop Routing table Built by the routing algorithm as a precursor to build the forwarding table Generally contains mapping from network numbers to next hops 154

155 Routing Routing table example: Forwarding table example: 155

156 Routing Network as a Graph Routing is a problem of graph theory. Networks can be represented as graphs Nodes(A-F) = hosts, switches, routers, or networks. (assume routers for now) The basic problem of routing is to find the lowest-cost path between any two nodes Where the cost of a path equals the sum of the costs of all the edges that make up the path 156

157 Routing For a simple network, we can calculate all shortest paths and load them into some nonvolatile storage on each node. Such a static approach has several shortcomings It does not deal with node or link failures It does not consider the addition of new nodes or links It implies that edge costs cannot change What is the solution? Need a distributed and dynamic protocol Two main classes of protocols for intra-domain routing: 1. Distance Vector 2. Link State 157

158 1. Distance Vector Each node constructs a one dimensional array (a vector) containing the distances (costs) to all other nodes and distributes that vector to its immediate neighbors Starting assumption is that each node knows the cost of the link to each of its directly connected neighbors Consider the following network graph: 158

159 Distance Vector Cost of each link = 1: Initial distances stored at each node (global view): View/Vector of A 159

160 Distance Vector STEP1: Each node set up an initial routing table Initial routing table at node A: A can reach B in one hop A cannot reach D 160

161 Distance Vector Step 2: immediate neighbors exchange routing table information Step3: nodes update their routing tables based on received routing table information E.g. FA : distance(f,g)=1 distance(a,g)= α >distance(f,g)+1=2 So A update its routing table: distance(f,g)= distance(a,f)+ distance(f,g) = 1+1 = 2 routing table at node A: distance(a,g)=1+1=2 distance(f,g)=1 2 distance(f,g)=1 Few such exchanges (repeat steps 2,3) are sufficient to build a complete routing table

162 Distance Vector Final routing table at node A 162

163 Distance Vector Final distances stored at each node (global view) 163

164 Distance Vector The distance vector routing algorithm is sometimes called as Bellman-Ford algorithm A node sends routing information to its neighbors under 2 circumstances: 1. Periodic update: Every T seconds each router sends its table to its neighbor(also let other nodes know its up and running) Receiver nodes update their routing table based on the new information T varies by protocol (seconds to minutes) 2. Triggered Update: Happens when a node notice a link failure 164

165 Distance Vector When a node detects a link failure: F detects that link to G has failed F sets distance to G to infinity and sends update to A A sets distance to G to infinity since it uses F to reach G A receives periodic update from C with 2-hop path to G A sets distance to G to 3 and sends update to F F decides it can reach G in 4 hops via A 165

166 Distance Vector F detects that link to G has failed: Distance(A,G) = α Distance(F,G) = α Distance(C,G) = 2 Distance(F,G) = α F sets distance to G to infinity and sends update to A Distance(F,G) = α A sets distance to G to infinity Distance(F,G) = α A receives periodic update from C with 2-hop path to G Distance(A,G) = 3 Distance(F,G) = 4 F sets distance to G 4 Distance(A,G) = 3 Distance(F,G) = α A sets distance to G to 3 and sends update to F 166

167 Routing Information Protocol (RIP) RIP is a routing protocol for exchanging routing table information between routers One of the more widely used routing protocols in IP networks Extremely simple Uses hop count as a path cost (link cost = 1). Therefore always tries to find minimal hop route Uses distance-vector algorithm Each router sends all or part of its routing table in routing updates. Periodic and triggered The updates are only sent to neighboring routers. 167

168 Routing Information Protocol (RIP) Routers advertise cost of reaching networks to their neighbors: Routers running RIP send their advertisements every 30 seconds Distance(C,2)=0 Distance(C,3)=0 Distance(C,5)=1 Distance(C,6)=1 Distance(C,4)=2 Example Network running RIP 168

169 Routing Information Protocol (RIP) Command Request or response. Version Specifies version of RIP. RIP version 1 used old classful addressing RIP version 2 support subnet mask Must be Zero Not used. Family of net 1 RIP Supports multiple address families, not just IP Type of address being specified (IP, IPX, etc) <Address,Mask,Distance> triples Routing table entries Takes majority of space of the packet RIPv2 Packet Format 169

170 2. Link State Routing Second type of intra-domain routing Strategy: Send to all nodes (not just neighbors) information(cost) about directly connected links (not entire routing table). All nodes can use this information to find shortest path to any node in the network Link State Routing protocols therefore rely on 2 mechanisms: 1. Reliable dissemination of link state information 2. Calculation of routes from the sum of all the accumulated link-state knowledge. 170

171 Link State Routing Link State Packet (LSP) id of the node that created the LSP cost of link to each directly connected neighbor sequence number (SEQNO) time-to-live(ttl) for this packet 1. Reliable Flooding Process of ensuring all nodes get a copy of link-state information from all the other nodes Link state information of a node is floodedwithin the network Generate new LSP periodically; increment SEQNO Store most recent LSP from each node (identified by SEQNO) ForwardLSP to all nodes but one that sent it Decrement TTLof each stored LSP; discardwhen TTL=0 171

172 Link State Routing Reliable Flooding (a) LSP arrives at node X (b) X floods LSP to A and C (c) A and C flood LSP to B (but not X) (d) flooding is complete Flooding of link-state packets 172