Introduction 7010INT Data Communications Lecture 7 The Layer Internetworking & Devices Connecting LANs Routing Backbone networks Virtual LANs Addressing Application Presentation Session Data Link Physical Addressing, Routing & Packet Delivery Text reference: Chapters 16, 19 & 21 Dr. Ruben Gonzalez School of Information Technology Connecting LANs Why Interconnect? Often you need to connect a local area network to another local area network or to a wide area network. Local area network-to-local area network connections are often performed with a bridge-like device. Local area network-to-wide area network connections are usually performed with a router. A third device, the switch, can be used to interconnect segments of a local area network. Application Gateway Presentation Session Router Data Link Bridge / Switch Physical Repeater / Hub To separate / connect one corporate division with another. To connect two LANs with different protocols. To connect a LAN to the Internet. To break a LAN into segments to relieve traffic congestion. To provide a security wall between two different types of users.
Repeaters Operate purely at the physical layer. Propagate all traffic in both directions between the LAN sections they link Used to extend the length of a LAN segment beyond the normal limit imposed by the medium. Regenerate all signals received on one segment and forwarding (repeats) them onto the next. They may connect different types of cable, but use the same data link and network protocol. LLC MAC PHY Repeater Repeater LLC MAC PHY Bridges A bridge (or bridge-like device) can be used to connect two similar LANs, such as two CSMA/CD LANs. A bridge can also be used to connect two closely similar LANs, such as a CSMA/CD LAN and a token ring LAN. The bridge examines the destination address in a frame and either forwards this frame onto the next LAN or does not. The bridge examines the source address in a frame and places this address in a routing table, to be used for future routing decisions. LAN Segment LAN Segment LAN Segment Bridges Transparent Bridge LLC MAC PHY LAN Segment Relay MAC MAC PHY PHY Bridge LAN Segment LLC MAC PHY A transparent bridge does not need programming but observes all traffic and builds routing tables from this observation. This observation is called backward learning. Each bridge has two connections (ports) and there is a routing table associated with each port. A bridge observes each packet that arrives at a port, extracts the source address from the packet, and places that address in the port s routing table. Transparent bridges are used with CSMA/CD LANs.
Transparent Bridge Transparent Bridge A transparent bridge can also convert one frame Header Frame contents format to another. 802.3 Traile Preamble SD DA SA Length DATA Pad FCS 802.4 Preamble Header SD Frame contents Trailer FC DA SA DATA FCS ED 802.5 Header Frame contents Trailer SD AC FC DA SA DATA FCS ED FS SD = Start Delimiter FC = Frame Control AC = Access Control DA = Destination address FCS = Frame check sequence ED = End Delimiter FS = Frame Status SA = Source address Note that some people / manufacturers call a bridge such as this a gateway or sometimes a router. The bridge removes the headers and trailers from one frame format and inserts (encapsulates) the headers and trailers for the second frame format. A Problem With Transparent Bridges A 1 Routing Table Dest Port A 1/2 D? A to D 2 3 Bridge Bridge 2 3 D Routing Table Dest Port A 1/2 D? F H
Spanning Tree Algorithm Spanning Tree Algorithm Initialisation Attempts to solve the routing problem Every port on every bridge is categorised as either a forwarding port or a blocking port Forwarding and blocking ports are assigned such that no loops can be formed Once the spanning tree has been established, any packets received by a bridge with an unknown destination are only sent out to on forwarding ports. Each bridge on the LAN is assigned an ID number (arbitrarily). The bridge with the smallest ID is chosen to be the root bridge by all themselves. Each port on every bridge is assigned a cost. The cost is determined by the bit rate of the attached segment: the higher the bit rate, the lower the cost. If all ports have the same bit rate then the cost is set to 1 for each port. Spanning Tree Algorithm - Operation Each bridge determines its root port (the port with the lowest path cost to the root). A designated port is found for each LAN segment. If only one bridge is connected to a LAN segment, then the designated port will be the port connecting that segment to the bridge. If more than one bridge is connected to a single LAN segment, then the designated bridge will be the one that can carry a frame from the LAN segment to the root with the cheapest path cost. The port that connects the designated bridge to the LAN is the designated port. Note that a root port cannot be selected as the designated port. Any ports that have not been assigned as either root ports or designated ports are labelled as blocking ports. Root ports and designated ports are labelled as forwarding ports. Path cost Segment 1 Segment 2 Root path Blocking port Segment 3 2 A Spanning Tree Des 200 2 Des RP BP RP 4 2 Segment 4 Segment 5 Root Des Des 4 100 4 2 500 300 400 2 2 RP Des RP BP Designated port 4 4
Source-routing Bridges A source-routing bridge is used in token ring networks. They do not learn from watching tables. When a workstation wants to send a frame, it must know the exact path of network / bridge / network / bridge / network If a workstation does not know the exact path, it sends out a discovery frame. The discovery frame makes its way to the final destination, then as it returns, it records the path. Used mostly with token ring LANs, so not generally used today. Using Bridges to Connect Different Types of LANs - Issues Frame format Each LAN protocol has its own frame format, hence conversion is necessary. Payload size The maximum allowable length for the data field varies between protocols. This may require a frame to be subdivided before being forwarded. (Eg from token ring to Ethernet). Data rate Different protocols operate at different data rates (as well as different LANs running the same protocol eg 10Base-T and 100Base-T). Frames travelling from a faster to a slower LAN must be buffered accordingly. Address bit order the binary representation of physical addresses varies between protocols. Appropriate conversion must be performed. Bridge Disadvantages Since a bridge receives and buffers all frames in their entirety before performing the forwarding them, it introduces an additional store-and-forward delay compared with a repeater. Bridges don t perform flow control and may overload during periods of high traffic. Bridging of segments operating with different data link protocols (and frame formats) means that the contents of frames must be modified prior to forwarding. This necessitates a new frame check sequence being generated by the bridge with the effect that any errors introduced while frames are being relayed across the bridge will go undetected.
Hubs Switches As seen earlier, a hub interconnects two or more workstations into a local area network. When a workstation transmits to a hub, the hub immediately resends the data frame out all connecting links. A hub can be managed or unmanaged. A managed hub possesses enough processing power that it can be managed from a remote location. A switch is a combination of a hub and a bridge. It can interconnect two or more workstations, but like a bridge, it observes traffic flow and learns. When a frame arrives at a switch, the switch examines the destination address and forwards the frame out the one necessary connection. Workstations that connect to a hub are on a shared segment. Workstations that connect to a switch are on a switched segment. Switches The backplane of a switch is fast enough to support multiple data transfers at one time. A switch that employs cut-through architecture is passing on the frame before the entire frame has arrived at the switch. Multiple workstations connected to a switch use dedicated segments. This is a very efficient way to isolate heavy users from the network. A switch can allow simultaneous access to multiple servers, or multiple simultaneous connections to a single server.
Switches v s Hubs Maximum total capacity = 10 Mbps 10 Mbps Hub 10 Mbps 10 Mbps 10 Mbps Maximum total capacity = N x10 Mbps Switch 10 Mbps 10 Mbps Routers Routers in a Connect two or more LANs with same network layers. Can determine which is the best path for a message to take between networks. A router can be a special purpose black box, a computer with several NICs or a special network software module within a computer. A router makes no change to the network layer packet it receives. It is quite common now for bridges to be bridge-routers and the distinction between the two is becoming blurred. Present Session DataLink PHY Router Router Router / Gateway Relay DataLink DataLink PHY PHY Router Router Present Session DataLink PHY 1 2
Routing Concepts Adaptive and nonadaptive routing In nonadaptive routing, once the best path to a particular destination has been determined, every single packet addressed to that destination is sent along that path. This happens regardless of any changes in network conditions. In adaptive routing, the routing tables are continually being updated to reflect changing network conditions (such as congestion, failure, reconfiguration, etc.). Hence every packet addressed to a particular destination may travel a different path even if some of the packets belong to a single message. Packet Lifetime When adaptive routing is used, it is possible that a packet can get lost in an infinite loop. To prevent this, an extra field to the packet called the packet lifetime or time to live (TTL). This field gives the number of hops that the packet can take before it is considered lost. Each router that receives the packet decrements the TTL field before forwarding the packet. If a router receives a packet with a TTL value of zero, it destroys the packet without forwarding it. Brouters Brouters are devices that combine the functions of both bridges and routers. They operate at both the data link & network layers. They can connects both same and different data link type network LAN segments. As fast as bridges for same data link type networks. Gateways (Protocol Converters) Gateway Illustration A gateway is a protocol converter. Used to connect LANs with same or different physical, data link and network layers. A gateway can be: A stand alone computer with special software and several NICs Software installed in a router. A front end processor (FEP) in a mainframe. A special circuit card installed in a network server. Some protocol conversions simply require a change in the header and trailer of the packet. Others are more complicated and can involve changing the data rate, packet size and overall packet format. SNA Gateway Netware
A Rose by Any Other Name ing Device Comparison The terminology used in the marketplace may differ substantially and tends to change as fast a catalogues can be printed! For example, one vendor s bridge may provide the functions of another s router. Some other networking devices that are available: Multiprotocol bridges translate between different data link layer protocols. Multiprotocol routers can understand several different network layer protocols. Protocol filtering bridges multiprotocol bridges that forward only packets of a certain type. Encapsulating bridges connect networks with different data link protocols. Layer-3 switches (IP switches) switch messages base on network layer address. Device OSI Layer Physical Data Link Hub/Repeater Physical Same/Different Same Same Bridge Data Link Same/Different Usually Same Same Switch Data Link Same/Different Same Same Routers Same/Different Same/Different Same Brouters /DL Same/Different Same/Different Same Gateway Application Same/Different Same/Different Same/Different Common Backbone Designs Routed Backbones When many LANs are connected together, the connecting network is usually called a backbone In designing/classifying backbone networks, the most important characteristic is the way in which packets are moved across the backbone. There are three basic backbone designs : routed, bridged and switched Routed backbones move packets based on their network layer address. The primary advantage is that routed backbones clearly segment each part of the network connected to the backbone. There are two disadvantages to routed backbones. The routers in the network impose time delays Routed networks require a lot of management.
Bridged Backbones Bridged/Routed Backbone Bridged backbones move packets based on their physical address. Advantages: Bridges tend to be less expensive than routers. Bridged backbones tend to be simpler to install. Disadvantages: Individual segment management is difficult. throughput is lower than routed backbones since broadcast messages must be permitted to travel everywhere leading to congestion. Hub LAN Backbone Switched Backbones Switched Backbone Most switched backbones use the data link layer address to move packets. A collapsed backbone is the most common form. Advantages: Improved performance Far fewer networking devices. Flexible very easy to install and move nodes. Potential disadvantages: More broadcast traffic Difficult to isolate and separately manage individual LANs. Use more cable and if the switch fails, so does the network. Hub Switch LAN Backbone
Virtual LANs Switches permit the creation of Virtual LANs (VLANs). These provide greater opportunities to manage the flow of traffic on the LAN to avoid congestion and enhance security. VLANs are groups of computers in an intelligent switched network. Intelligent Switches Intelligent switches can be connected together to create very large networks supporting hundreds of network ports. Since there is not enough capacity in the backplane to support all ports if they become active the switch forms groups of connections and assigns capacity using time division multiplexing. This means that the switch no longer guarantees simultaneous transmission on all ports, but will accept simultaneous input and will switch incoming data to outgoing ports as fast as possible. These groups are called VLANs Port and MAC Based VLANs Port-Based VLANs (Layer-1 VLANs) Use the physical layer port address to form the VLAN groups. It is logical to connect computers that are physically close together on the LAN into ports that are physically close together on the switch, and to assign ports that are physically close together into the same VLAN. This is the traditional approach in LAN design: physical location determines the LAN, but is not always the most effective approach. Ports can be used to balance capacity against traffic MAC-Based VLANs (Layer-2 VLANs) MAC-based VLANs use the dame data link layer addresses to form the VLAN groups. The advantage is that they are simpler to manage when computers are moved. IP and Application Based VLANs IP-Based VLANs(Layer-3 VLANs) IP-based VLANs use the network layer address (i.e. TCP/IP address) to form the VLAN groups. Layer-3 VLANs reduce the time spent reconfiguring the network when a computer is moved as well. Some layer-3 VLANs can also use the network layer protocol to create VLAN groups. This flexibility enables manager even greater precision in the allocation of network capacity. Application-Based VLANs (Layer-4 VLANs) Application-based VLANs use the application layer protocol in combination with the data link layer and network layer addresses to form the VLAN groups. The advantage is a very precise allocation of network capacity.
TCP/IP Layer (IP) IP (Internetwork Protocol) is an unreliable, connectionless protocol. The data units (packets) in IP are called datagrams and have the following format: Addressing In order for messages to get to the correct destination, addresses must be specified. Three main levels of addressing are used in a TCP/IP network: Header Format: 1 st byte in header Service VER HLEN Type Identification Time to live Protocol Source IP address Destination IP address Options Total length Flags Fragmentation Off (13 bi ) Header Checksum Datagram Format: Header 20-60 bytes Data 20-65536 bytes Address Type Application Layer Layer (Logical) Data Link Layer (Physical) Eg of Protocol/Software Web browser / email client Example www.int.gu.edu.au / someone@gu.edu.au TCP/IP 123.234.112.9 Ethernet 00-0C-00-F5-03-5A 1 byte (8 bits) IP Addresses Dotted Decimal Notation In TCP/IP the network layer address is called the IP address Five classes of addresses are defined to cater for the needs of various organisations The format of IP addresses is as follows: Byte 1 Byte 2 Byte 3 Byte 4 Class B 10000000 Netid 00001011 00000011 Hostid 00011111 128.11.3.31 Byte 1 Byte 2 Byte 3 Byte 4 Class A Class B Class C Class D Class E 0 Netid Hostid 10 Netid 110 Netid 1110 Multicast address 1111 Reserved Hostid Hostid Class A B C Available Addresses 16 Million 16,000 250 Address Structure 50.x.x.x 128.192.x.x 192.11.56.x NetID (bits) 7 14 21 HostID (bits) 24 16 8
TCP/IP Internet Addressing Example Subnet Addressing Note that devices with more than one network connection have more than one network (IP) address. 220.3.6.5 220.3.6.7 220.3.6.0 220.3.6.1 129.8.0.1 129.8.9.14 134.18.0.0 129.8.0.0 129.8.7.15 134.18.1.29 134.18.0.209 222.13.16.12 222.13.16.7 G G 222.13.16.0 134.18.0.2 220.3.6.9 134.18.0.210 a.b.c.d To Internet R 134.18.5.9 207.42.56.10 207.42.56.0 207.42.56.11 R Outside world Netid Hosttid 128.11.3.31 Netid Hosttid Subnetid Within Organisation 124.0.0.0 124.0.132.119 124.1.0.5 124.1.1.15 124.1.1.12 124.0.12.2 Subnetwork Masks Subnetworking Example A network node identifies the subnetwork portion of the address using a mask. A mask is a 32-bit pattern that is ANDed with an IP address. A subnetwork mask will have a 1 for all bits that indicate the subnet, and 0 for all bits that designate the host. Examples, The subnetwork mask for the case in the previous eg would be 255.255.255.0 The subnetwork mask for an organisation that has been assigned a class C network address and wishes to use the first three bit of the last byte to designate the subnet would be: 255.255.255.224 141.14.2.9 141.14.2.1 141.14.2.0 subnetwork R 141.14.4.0 subnetwork 141.14.12.0 141.14.12.10 141.14.12.10 141.14.12.12 141.14.2.10 141.14.12.109 141.14.4.11 subnetwork : 141.14.0.0 141.14.4.1 141.14.4.101 141.14.4.140
Other Protocols in the TPC/IP Layer Address Resolution Protocol Address Resolution Protocol (ARP) associates an IP (network) address with the physical address (the address on the NIC). Internet Control Message Protocol (ICMP) used by hosts and routers to inform sender of problems with datagram delivery. Internet Group Message Protocol (IGMP) used to support multicasting. Used by a router to determine which hosts on a LAN belong to the multicast group defined by a given class D address. I m looking for the physical address of the node whose IP address is 141.14.22.1 ARP Request I m the node you re after and my physical address C0-12-AF-BG-33-01 ARP Response