10-Mbps and 100-Mbps Ethernet 319 Manchester encoding relies on the direction of the edge transition in the middle of the timing window to determine the binary value for that bit period. In the encoding exam- ple shown in Figure 6-3, one timing window is highlighted vertically through all four waveform examples. The top waveform has a falling edge in the center of the timing window, so it is interpreted as a binary 0. The result is that in the center of the timing window for the second waveform, there is a rising edge, which is interpreted as a binary 1. Instead of a repeating sequence of the same binary value in the third waveform example, there is an alternating binary sequence. In the first two examples, the signal must tran- sition back between each bit period so that it can make the same-direction transition each time in the center of the timing window. With alternating binary data, there is no need to return to the previous voltage level in preparation for the next edge in the center of the timing window. Thus, any time there is a long separation between one edge and the next, you can be certain that both edges represent the middle of a timing window. The fourth waveform example is random data that enables you verify that whenever there is a wide separation between two transitions, both edges are in the center of a timing window and represent the binary value for that timing window. Legacy (10-Mbps) Ethernet has some common architectural features. All of these legacy versions are referred to as shared Ethernet because they share a common collision domain. It is not only allowed, but it is expected that an Ethernet network could contain multi- ple types of media (for example, 10BASE5, 10BASE2, 10BASE-T, and so on). The standard goes out of its way to ensure that interoperability is maintained. However, when implementing a mixed-media network, it is important to pay particular attention to the overall architecture design. It becomes easier to violate maximum delay limits as the network grows and becomes more complex. The timing limits are based on param- eters such as these: ■ Cable length and its propagation delay ■ Delay of repeaters ■ Delay of transceivers (including NICs, hubs, and switches) ■ Interframe gap shrinkage ■ Delays within the station Lab Activity Waveform Decoding The purpose of this lab is to integrate knowledge of networking media; OSI Layers 1, 2, and 3; and Ethernet, by decoding a digital waveform of an Ethernet frame. 1102.book Page 319 Tuesday, May 20, 2003 2:53 PM 320 Chapter 6: Ethernet Technologies and Ethernet Switching 5-4-3 Rule 10-Mbps Ethernet operates within the timing limits offered by a series of no more than five segments separated by no more than four repeaters. That is, no more than four repeaters can be connected in series between any two distant stations. The coaxial implementations have a further requirement that there can be no more than three pop- ulated segments between any two distant stations. The other two allowed coaxial seg- ments are used to extend the diameter of the collision domain and are called link segments. The primary characteristic of a link segment is that it has exactly two devices attached. All twisted-pair links, such as 10BASE-T, meet the definition of a link segment. 10BASE5 The original (1980) Ethernet product (10BASE5) transmitted 10 Mbps over a single thick coaxial cable bus, thus the name Thicknet. 10BASE5 is important for historical reasons: It was the first medium used for Ethernet. 10BASE5 was part of the original 802.3 standard. It can be found today as part of legacy installations. It is not a preferred choice for new networks because its primary benefit, length, can be accomplished in other ways. Although 10BASE5 systems are inexpensive and require no configuration (there is no need for hubs to extend the length of the system), basic components such as NICs are very difficult to find, and the technology is very sensitive to signal reflections on the cable. In addition, 10BASE5 systems are very cable-dependent across the whole collision domain and thus represent a large single point of failure. The timing, frame format, and transmission process were described previously in Chapter 5, “Ethernet Fundamentals,” and are common to all 10-Mbps legacy Ethernet. 10BASE5 uses Manchester-encoded signals on thick coaxial cable. Figure 6-4 is an example of a 10BASE5 signal. It is transmitted from approximately 0V to –1V. 10BASE5 potentially could be idle (0V) for days if no station wanted to transmit. 10BASE5 is asynchronous. Figure 6-4 10BASE5 Signal Decoded 1102.book Page 320 Tuesday, May 20, 2003 2:53 PM 10-Mbps and 100-Mbps Ethernet 321 In Figure 6-4, timing marks have been added to aid you in recognizing the timing windows from which the binary data was decoded. The y-axis is voltage; the x-axis is time. Voltage has been measured between the central conductor and the outer sheath- ing of the coaxial cable. A 10BASE5 thick coax cable, as shown in Figure 6-5, has a solid central conductor, a minimum nominal velocity of propagation (NVP) of 0.77c, and 50 ohms of impedance/ termination resistance; it uses N-style screw-on connections. Each of the maximum five segments of thick coax can be up to 500m (1640 ft.) long, and each station is con- nected to a transceiver on the coax via an Attachment Unit Interface (AUI) cable that can be up to 50m (164 ft.) long. The cable is large, heavy, and difficult to install, but the distance limitations were favorable; this prolonged its use in certain applications. Figure 6-5 10BASE5 Thicknet Cable Other specifications or limitations of 10BASE5 cable include the following: ■ Only one station can transmit at a time (or a collision will occur). ■ 10BASE5 can run only in half-duplex mode, subject to the CSMA/CD rules. ■ Up to 100 stations, including repeaters, can exist on any individual 10BASE5 segment. 10BASE2 10BASE2 (originally 802.3a-1985) was introduced in 1985 because its coaxial cable of a smaller size, lighter weight, and greater flexibility made installation easier than 10BASE5. Because of its use of thinner cable, 10BASE2 often is referred to as Thinnet. 10BASE2 still exists in legacy networks. Although there is little reason to install a 10BASE2 net- work today, its low cost and lack of need for hubs are attractive. Essentially, 10BASE2 requires no configuration, although obtaining NICs is increasingly difficult. Just like 10BASE5 systems, 10BASE2 systems are very cable-dependent across the whole colli- sion domain and represent a large single point of failure. The timing, frame format, and transmission were described previously in Chapter 5 and are common to all 10-Mbps Legacy Ethernet. 1102.book Page 321 Tuesday, May 20, 2003 2:53 PM 322 Chapter 6: Ethernet Technologies and Ethernet Switching 10BASE2 uses Manchester-encoded signals on thin coaxial cable. A 10BASE2 signal is transmitted from approximately 0V to –1V. (The y-axis is voltage; the x-axis is time. Voltage is measured between the center conductor and the outer sheathing conductor.) 10BASE2 potentially could be idle (0V) for days if no station wanted to transmit. 10BASE2 is asynchronous. The computers on the LAN were linked together like the beads of a necklace by an unbroken series of coaxial cable lengths. These lengths of coaxial cable were attached by British Naval Connectors (BNCs) to a T-shape connector on the NIC, as shown in Figure 6-6. This single coaxial cable was the shared bus for the network. Workstations easily could be moved and reattached, or new workstations could be added to the LAN. Otherwise, 10BASE2 used the same original Ethernet half-duplex protocol. A 10BASE2 thin coax cable, as shown in Figure 6-8, has a stranded central conductor. (Be sure that stranded coax is specified when new cable is ordered. Some installers find it hard to work with and use solid-core coax when possible.) It has a minimum nomi- nal velocity of propagation (NVP) of 0.65c, has 50 ohms of impedance/termination resistance, and uses BNC T-style connections. Each of the maximum five segments of thin coax can be up to 185m long (600 ft.), and each station is connected directly to the BNC T connector on the coax. Figure 6-6 Thinnet and BNC Connector 10BASE-T 10BASE-T (originally 802.3i-1990) substituted the cheaper and easier-to-install UTP copper cable for coaxial cable. This cable plugged into a central connection device, a hub or a switch, that contained the shared bus. The type of cable used in 10BASE-T, the distances that the cable could extend from the hub, and the way in which the UTP was installed, interconnected, and tested were standardized in a “structured cabling 1102.book Page 322 Tuesday, May 20, 2003 2:53 PM 10-Mbps and 100-Mbps Ethernet 323 system,” which increasingly specified a star or extended star topology. 10BASE-T was originally a half-duplex protocol, but full-duplex features were added later. The explo- sion in Ethernet’s popularity in the 1990s—when Ethernet came to dominate LAN technology—was 10BASE-T running on Category (Cat) 5 UTP. To reacquaint yourself with network topologies and networking media, refer back to Chapter 2, “Networking Fundamentals,” and Chapter 3, “Networking Media.” The timing, frame format, and transmission were described previously and are common to all 10-Mbps legacy Ethernet. 10BASE-T uses Manchester line-encoded signals over Category 3 (now 5, 5e, or better) UTP. 10-Mbps Ethernet is asynchronous, and the cable often is completely idle (0V) for long periods of time between transmissions. 10BASE-T links have a link pulse present about every 125 milliseconds (eight times per second), but can otherwise be idle. 10BASE-T networks are “alive” with link pulses. A 10BASE-T unshielded twisted-pair (UTP) cable has a solid conductor for each wire in the maximum 90m horizontal cable, which should be 0.4 mm to 0.6 mm (26 to 22 American Wire Gauge [AWG]) in diameter. The 10m of allowed patch cables use similar- dimension stranded cable for durability because it is expected to experience repeated flexing. Suitable UTP cable has a minimum NVP of 0.585c, has 100 ohms of impedance, and uses eight-pin RJ-45 modular connectors as specified in ISO/IEC 8877. Cables between a station and a hub generally are described as between 0m and 100m long (0 ft. to 328 ft.), although the precise maximum length is determined by propagation delay through the link segment (any length that does not exceed 1000 ns of delay is acceptable). Usually, 0.5 mm (24 AWG) diameter UTP wire in a multipair cable will meet the requirements at 100m. Although Category 3 cable is adequate for use on 10BASE-T networks, it is strongly recommended that any new cable installations be made with Category 5e or better materials and wiring practices. Use all four pairs, and use either the T568A or T568B cable pinout arrangement. With this type of cable installation, it should be possible to operate many different media access protocols (including 1000BASE-T) over the same cable plant, without rewiring. 1102.book Page 323 Tuesday, May 20, 2003 2:53 PM 324 Chapter 6: Ethernet Technologies and Ethernet Switching Table 6-2 shows the pinout for a 10BASE-T connection. Notice that two separate transmit/receive paths exists (whereas coaxial cable has only one). Figure 6-7 shows conceptual and physical connections between two stations. A cross- over cable is required, so Tx on device A sends signals to Rx on device B. Note that two point-to-point connections exist (TxA to RxB, and TxB to RxA). Figure 6-7 10BASE-T Station to Station Table 6-2 10BASE-T Cable Pinouts Pin Number Signal 1 TD+ (Transmit Data, positive-going differential signal) 2 TD– (Transmit Data, negative-going differential signal) 3 RD+ (Receive Data, positive-going differential signal) 4 Unused 5 Unused 6 RD– (Receive Data, negative-going differential signal) 7 Unused 8 Unused 1102.book Page 324 Tuesday, May 20, 2003 2:53 PM 10-Mbps and 100-Mbps Ethernet 325 Figure 6-8 shows the connection between stations and repeaters, multiport repeaters (hubs), or switches. The same connection would be used between a router and a hub or a switch. A straight-through cable is used. Note that inside the hub is a bus topol- ogy, which is a collision domain. When a workstation is connected to a switch using a straight-through cable, all individual links are point-to-point. The switch fabric circuitry allows full bandwidth simultaneously between pairs of ports without collisions. Figure 6-8 10BASE-T Straight-Through Cable Because station-to-station, switch-to-switch, and station-to-switch connections all are point-to-point links, they have two physically separate communication pathways/ channels on two separate UTP wire pairs. In this case, collisions are not physical events, but rather the result of the decision to not allow simultaneous Tx and Rx. Thus, either half duplex (subject to the administrative imposition of CSMA/CD) or full duplex (no physical collisions occur) is a configuration choice. Most of the time, you run these connections in full duplex, which not only eliminates collisions, but also doubles the throughput of the connection. When first introduced, the relevant IEEE standard was entitled 802.3x-1997 Full-Duplex. However, station-to-hub connections involve the bus topology within the hub, an actual physical collision domain. Hence, this connec- tion can run only half duplex and is subject to CSMA/CD because of the physical nature of the structure. 10BASE-T carries 10 Mbps of traffic in half-duplex mode; however, 10BASE-T in full- duplex mode actually can exchange 20 Mbps of traffic (although, again, some of this is overhead, not user data). This concept will become increasingly important with the desire to increase the speed of Ethernet links. 1102.book Page 325 Tuesday, May 20, 2003 2:53 PM 326 Chapter 6: Ethernet Technologies and Ethernet Switching 10BASE-T Architecture 10BASE-T links generally consist of a connection between the station and a hub or switch. Hubs should be thought of as multiport repeaters and count toward the limit on repeaters between distant stations. Switches can be thought of as multiport bridges and are subject to 100m length limitations but no limit on switches between distant stations. Although hubs can be linked in series (sometimes called daisy-chaining, or cascading), it is best to avoid this arrangement when possible, to keep from violating the limit for maximum delay between distant stations. The physical size of a 10BASE-T network is subject to the same rules as 10BASE5 and 10BASE2 concerning the number of repeaters. When multiple hubs are required, it is best to arrange them in hierarchical order, to create a tree structure instead of a chain. Also, performance will be improved if fewer repeaters separate stations. “Stackable” hubs, or concentrators with common backplanes that will support several multiport adapter cards, permit large numbers of stations to be connected to a device that counts as a single hub (repeater). Daisy-chaining switches is fine and is not subject to restrictions. All distances between stations are acceptable, although in one direction, the architecture is at its limit. The most important aspect to consider is how to keep the delay between distant stations to a minimum—regardless of the architecture and media types involved. A shorter maximum delay provides better overall performance. Consider the following architectures: ■ In Figure 6-9, there are five segments and four repeaters from Station 1 to any other station in these paths. For 10BASE-T connections, the maximum of three segments with stations does not apply because no other stations are on the same cable. Each connection is described as a link segment. Figure 6-9 Example 10-Mbps Mixed Architecture 1 1102.book Page 326 Tuesday, May 20, 2003 2:53 PM 10-Mbps and 100-Mbps Ethernet 327 ■ In Figure 6-10, from any station (except Station 1) to any other station, the path is only three repeaters. Because these alternate paths include 10BASE5 and 10BASE2 links, the other requirements still apply there (such as only three seg- ments with stations). Figure 6-10 Example 10-Mbps Mixed Architecture 2 10BASE-T links can have unrepeated distances up to 100m. This might seem like a long distance, but it typically is used up quickly when wiring an actual building. Hubs can solve this distance issue, although a maximum of four repeaters could be chained together because of timing considerations. The widespread introduction of switches has made this distance limitation less important. As long as workstations are located within 100m of a switch, the 100m distance starts over at the switch, which could be connected via another 100m to another switch, and so on. Because most modern 10BASE-T Ethernet is switched, these are the practical limits between devices. Ring, star, and extended star topologies all are allowed. The issue then becomes one of logi- cal topology and data flow, not timing or distance limitations. Table 6-3 shows a chart of the 10BASE-T link characteristics. Table 6-3 10BASE-T Link Characteristics Chart Connection Maximum Segment Station to station, station to switch, switch to switch 100m, with no limitations on daisy chaining Station to hub 100m, but subject to four-repeater rule 1102.book Page 327 Tuesday, May 20, 2003 2:53 PM 328 Chapter 6: Ethernet Technologies and Ethernet Switching 100-Mbps Versions of Ethernet 100-Mbps Ethernet, also known as Fast Ethernet (in comparison to the original 10-Mbps Ethernet), was a series of technologies. The two technologies that became commercially important are 100BASE-TX (copper UTP-based) and 100BASE-FX (multimode optical fiber-based). This section examines the commonalities between these two technologies and then examines their differences individually. Three things are common to 100BASE-TX and 100BASE-FX: ■ The timing parameters ■ The frame format ■ Parts of the transmission process Table 6-4 shows the parameters for 100-Mbps Ethernet operation. 100BASE-TX and 100BASE-FX both share timing parameters. Note that 1 bit-time in 1000-Mbps Ethernet is 10 nsec = .01 microseconds = 1 100-millionth of a second. The 100-Mbps frame format is the same as the 10-Mbps frame. Unlike 10-Mbps Ethernet, in which the process was the same for all technologies until the signal was applied to the medium. Fast Ethernet represents a tenfold increase in speed. With this increase in speed comes extra requirements. The bits being sent get shorter in duration and occur more frequently. They require more careful timing, and their transmission requires frequencies closer to Table 6-4 Parameters for 100-Mbps Ethernet Operation Parameter Value Bit-time 10 nsec Slot time 512 bit-times Interframe spacing 96 bits Collision attempt limit 16 Collision backoff limit 10 Collision jam size 32 bits Maximum untagged frame size 1518 octets Minimum frame size 512 bits (64 octets) 1102.book Page 328 Tuesday, May 20, 2003 2:53 PM . in Chapter 5 and are common to all 10 -Mbps Legacy Ethernet. 11 02. book Page 3 21 Tuesday, May 20 , 20 03 2: 53 PM 322 Chapter 6: Ethernet Technologies and Ethernet Switching 10 BASE2 uses Manchester-encoded. segment. Figure 6-9 Example 10 -Mbps Mixed Architecture 1 110 2. book Page 326 Tuesday, May 20 , 20 03 2: 53 PM 10 -Mbps and 10 0-Mbps Ethernet 327 ■ In Figure 6 -10 , from any station (except Station 1) to any other. integrate knowledge of networking media; OSI Layers 1, 2, and 3; and Ethernet, by decoding a digital waveform of an Ethernet frame. 11 02. book Page 319 Tuesday, May 20 , 20 03 2: 53 PM 320 Chapter 6: Ethernet