Published on December 30, 2007
Slide1: Today’s Lecture – Connecting LANs LAN Connecting Devices Bridges Transparent Bridges High speed LANs Installing a LAN Slide2: LAN Connecting Devices Repeater propagate the signal from one segment to another. Hub (MultiWay Repeater) propagate the signal onto a number of segments, possibly with different types of cable. Bridge connect two local area networks together. Bridges can be used to localize traffic. Repeater Layout: Repeater Layout ILLEGAL Legal Slide4: Repeaters: used in Ethernet (802.3) There is a distance limitation on Ethernets (802.3) because the electrical signal becomes weaker as it travels along a wire. To overcome this limitation, a repeater can be used to allow two cables to be joined together. A repeater is an analog device. When a repeater sense a signal on one cable, the repeater transmits an amplified copy on the other cable. Repeaters do not understand the frame format, they only deals with electrical signals. The maximum length of an Ethernet cable is 500 metres (10Base2 = 200m, 10BaseT = 100m) Slide5: Repeaters: used in Ethernet (802.3) The length of a Ethernet cannot be increased indefinitely by using multiple repeaters. The reason for this is because Ethernet uses Carrier Sense Multiple Access with Collision Detection (CSMA/CD). If the time it takes for a frame to propagate from one end of the network to the other end increases, it reduces the possibility of the computers connected to the cable being able to detect collisions. Ethernet (802.3) standard specifies that the network may not operate correctly if more than four repeaters separate any two computers. Slide6: Hub Hub (MultiWay Repeater) propagate the signal onto a number of segments, possibly with different types of cable. A hub consists of a single box with multiple ports. Each computer has a 10BaseT cable connecting to a port on the hub. A signal coming into the hub on a particular port is propagated to all other ports. Thus the hub simulates a single shared medium. Slide7: Bridges Bridges are used to connect two or more LANs together into one big LAN. Bridges work at the Data Link Layer. (They do not look at the Network Layer header as in the case of routers). Why do we need bridges? LAN’s in different departments may have developed independently. Eventually they may need to be connected. Organisations may have LANs in different buildings, a bridge is needed to connect them. If a LAN is very busy, it may need to be split into a number of LANs to handle the load better (reduce collisions). Slide8: Bridges Bridges are sometimes used to partition a network to contain damage. If a computer has a defective Network Interface Card, a bridge would not forward defective frames from one cable to the other cable (unlike a repeater). Bridges increase the possibility of reliability in the network. Bridges are good for security, as they can be programmed to avoid forwarding sensitive traffic. Slide9: Diagram of a bridge Data Link Network MAC Physical CSMA/CD LAN Token Ring LAN Slide10: Explanation of diagram Host A has a packet to send The packet is passed to the Data Link Layer and gets a sequence number etc. This is then passed to the MAC sub-layer, which adds information relevant to 802.3. A frame is created at this point. The frame is passed to the physical layer, which places it on the cable as a sequence of 1’s and 0’s. The frame arrives at the bridge and is eventually passed up to the MAC sub-layer where the 802.3 specific information is stripped off. The bare packet is handed to the Data Link Layer. Slide11: Explanation of diagram The Data Link Layer sends the packet down a different protocol stack, in this case 802.5. The frame is eventually sent to the other LAN. Note: It is important to remember that a bridge connecting different types of Local Area Networks will have: Different MAC sub-layers for each type of LAN Possibly different physical layers Slide12: Transparent Bridges Transparent Bridges are an IEEE standard for bridges. If you have a site with multiple LANs, you can plug these LANs into a Transparent Bridge and everything should work. There should be no need to make hardware or software changes, no address setting, no manual building of routing tables. The Transparent Bridge should operate using ‘plug and play’ The most important part of Transparent Bridges is its automatic building of routing tables, using the Backward Learning Algorithm. Slide13: Backward Learning Algorithm When Transparent Bridges are first plugged in all the routing tables are empty. The Transparent bridges need to build these routing tables as quickly as possible. The routing table consists of addresses of computers and the LAN that they are connected to. Every frame they receive, for which the Transparent Bridge does not have a destination address in their routing table, are forwarded to all LANs except the one that it comes in on. This is known as flooding the network. Slide14: Backward Learning Algorithm Before the Transparent Bridge forwards the frame, it makes an entry in its routing table for the computer sending the frame (source address in the frame). As part of this routing table entry, the Transparent Bridge notes the position of the sending computer, i.e. what LAN the sending computer is connected to. It uses this information for future reference, if another computer is sending frames to that particular computer. As more frames are received by the Transparent Bridge, the routing table at the Transparent Bridge fills up. As the routing table increases in size, the need to carry out flooding is reduced. Slide15: Backward Learning Algorithm The layout of the network can change as computers and bridges are powered up and down and moved around. All entries in the routing table are time stamped, if a frame arrives and its destination address is already in the routing table, the timestamp is updated. Every so often tables are purged of their entries (after a couple of minutes). This could happen when a machine is unplugged from the LAN and moved to another LAN. This algorithm means that machines that do not transmit regularly will get their table entries deleted and new entries in the routing table will have to be created when frames are being sent again. Slide16: Example Transparent Bridge B2’s routing table would list the address of host A as being on LAN 2. It doesn’t need to know where it is on LAN 2 LAN 1 LAN 2 LAN 3 LAN 4 Slide17: High Speed LANs The LANs that have been covered so far are mainly low speed and good only over short distances. The future of LANs lie with faster transmission over greater distances and with greater reliability. Two high speed LAN technologies of interest are FDDI: Fiber Distributed Data Interface, based on Optical Fiber Fast Ethernet, 802.3u, based on Twisted Pair and Optical Fiber. Slide18: FDDI: Fiber Distributed Data Interface. FDDI is a high speed token ring fiber optical cable based technology. This technology can ‘self-heal’, which means that the hardware can automatically detect and correct hardware problems. FDDI consists of two independent rings to connect each computer. Data flows in opposite direction in each ring. FDDI can have two types of Network Interface Cards, A and B, that connect to it. Class A Network Interface Cards connect to both rings while class B Network Interface Cards connect to only one ring. Slide19: FDDI: Fiber Distributed Data Interface. Only class A cards can be used to heal broken rings. Thus the number of class A cards cards define the fault tolerant characteristics of the network. When an error occurs the nearest computer routes frames from the inner ring to the outer ring. Slide20: FDDI Characteristics Twisted pair cabling can be used. LED are used instead of lasers FDDI was supposed to be the LAN of the future. The reason it has not become widespread is that the technology is complicated and hence the hardware is expensive Slide21: Fast Ethernet 802.3 The IEEE enhanced the old Ethernet standard to 100Mbsec LAN called Fast Ethernet. Some of the reasons that the IEEE developed this standard are: The need for a High Speed LAN standard to be backward compatible with existing LANs. Possible unforeseen problems of a newly developed protocol. The basic idea behind Fast Ethernet was simple: Keep all the old packet formats but reduce the time it takes to transmit a bit from 100 nanoseconds to 10 nanoseconds. The wiring is based on 10BaseT, 10Base5 and 10Base2 cannot be used. Thus Fast Ethernet uses hubs. Vampire Taps or BNC connectors cannot be used. Slide22: Fast Ethernet 802.3 Two choices of 10BaseT (Twisted pair) cables. Category 3 and Category 5. The advantage of Category 3 cable is that it gained widespread acceptance in offices throughout America and Europe. Therefore buildings would not have to be re-wired. The disadvantage of using Category 3 was its inability to carry the signals 100 metres as specified by 10BaseT. Category 5 twisted pair could carry signals 100 metres, as can fiber optic cable. The compromise chosen was to allow all three cabling possibilities. Slide23: Fast Ethernet Cabling Standards 100BaseT4: Twisted Pair - requires 4 Category 3 UTP cables; 100m max length. 100BaseTX: Twisted Pair - requires 2 Category 5 UTP cables; 100m max length 100BaseFX: Two strands of Multimode Fiber (one for each direction); 2km max length (distance between station and hub). Slide24: Installing a LAN Decide on the cabling and the network card that you will be using. Typically 10Base2 or 10BaseT is chosen because it is cheap and connections are reliable and easy to install. Ethernet (802.3) is cheap and easy to install. Select the type of workstations and file servers that will run the applications. Each workstation and file server requires a network interface card. In order to talk to the card, the workstation or file server needs a network interface card driver. File servers are just computers with a fast processor, large disks and memory. File servers can store users private storage area, databases and shared storage. Slide25: Installing a LAN Information on file servers are typically backed up to a tape archive in case the file server crashes. This limits the amount of important information that could be potentially lost. Decide on whether a printer is needed and whether it will be shared by a number of users connected to the network. Once all the above has been decided, network interface cards can be installed in each machine and the machines can be connected to the cable. The networking software on each workstation and file server has then to be configured.