CCNA 1 and 2 Companion Guide, Revised (Cisco Networking Academy Program) part 37 potx

10-Mbps and 100-Mbps Ethernet 329
medium-bandwidth limitations and become more susceptible to noise. In response to
these issues of synchronization, bandwidth, and signal-to-noise ratio (SNR), two sepa-
rate encoding steps are used by 100-Mbps Ethernet. The basic idea is to use codes—
which can be engineered to have desirable properties—to represent the user data in a
way that is efficient to transmit, including synchronization, efficient usage of band-
width, and improved SNR characteristics. The first part of the encoding uses a technique
called 4-bit/5-bit (4B/5B); the second part of the encoding is the actual line encoding
specific to copper or fiber.
The two forms of 100-Mbps Ethernet of consideration in this course, 100BASE-TX
and 100BASE-FX, encode nibbles (4-bit groupings) from the upper parts of the MAC
sublayer. The 4-bit patterns are converted into 5-bit symbols; symbols sometimes con-
trol information (such as start frame, end frame, or medium-is-idle conditions). The
entire frame to be transmitted is comprised of control symbols and data symbols (data
code groups). Again, all of this extra complexity is necessary to achieve the tenfold
increase in network speed.
After the 4B/5B encoding, the bits (in the form of code groups) still need to be placed
on the medium (that is, they must be line-encoded). The conversion from 4 bits to
5 bits also means that there are now 125 Mbps to be transmitted instead of 100 Mbps
during the same time interval. This puts more strict requirements on the medium,
transmitters, and receivers used. For example, the cable must be tested at higher fre-
quencies to ensure proper transmission characteristics. Any time that there is no data
to be sent, “idle code” groups still are sent to fill the empty periods and maintain syn-
chronization. At this point, the data path diverges depending on whether you are using
copper (100BASE-TX) or fiber (100BASE-FX) media.
100BASE-TX
The need for faster networks led to the announcement of the 100BASE-T Fast Ethernet
and autonegotiation standard in 1995 (originally 802.3u-1995). 100BASE-T increased
Ethernet’s bit rate to 100 Mbps. 100BASE-TX was the Category 5 UTP version of
100BASE-T that became commercially successful. Soon 10/100 hubs and switches
enabled Ethernet transmissions at the original rate of 10 Mbps to share the network

with frames sent at 100 Mbps.
The original coaxial Ethernet used half-duplex transmission; therefore, only one device
could transmit at a time. In 1997, Ethernet was expanded to include a full-duplex
capability (originally 802.3x) that allowed more than one PC on a network to transmit
at the same time. Devices called Ethernet switches were developed that enabled this
full-duplex communication and handled network traffic more efficiently than hubs.
1102.book Page 329 Tuesday, May 20, 2003 2:53 PM
330 Chapter 6: Ethernet Technologies and Ethernet Switching
These switches increasingly replaced hubs in high-speed networks because of their full-
duplex capability and rapid handling of Ethernet frames.
The timing, frame format, and transmission were described previously in Chapter 5
and are common to both versions of 100-Mpbs Fast Ethernet considered in this chapter.
100BASE-TX uses 4B/5B encoded data, which then is scrambled and converted to
multilevel transmit—three levels, or MLT-3, line encoding on Category 5 UTP (or
better). The MLT-3 encoding converts the binary data stream to an electrical wave-
form using a continuous signaling system. MLT-3 is different from nonreturn to zero
inverted (NRZI), in that the signal level alternates between above and below the zero
level instead of using only two levels.
Figure 6-11 shows some MLT-3 encoding examples. The basic rule of MLT-3 is that
binary 1s cause the voltage level to cycle to the next level down and then back up again.
Binary 0s do not cause a level transition.
Figure 6-11 MLT-3 Encoding Examples
In the encoding example in Figure 6-11, one timing window is highlighted vertically
through all four waveform examples. The top waveform has no transition in the center
of the timing window. No transition indicates that a binary 0 is present. If the example
waveform was all 0s on that line, the signal level represented would be either a constant
high, 0, or low across the waveform. A single 1 was introduced to move the remaining
0s to a different voltage level and show that 0s can appear at more than one voltage
level. The level depends upon what the previous (high or low) voltage level was and
moves in the opposite direction. In this first waveform, it is thus evident that the previ-

ous voltage level (not shown) was a low level.
N
O
TE
In NRZI encoding,
signals maintain con-
stant voltage levels
with no signal transi-
tions (no return to a
0V level) but interpret
the presence of data at
the beginning of a bit
interval as a signal
transition. Likewise,
they interpret the
absence of data as
no transition.
1102.book Page 330 Tuesday, May 20, 2003 2:53 PM
10-Mbps and 100-Mbps Ethernet 331
The second waveform in Figure 6-11 has a transition in the center of the timing window.
A binary 1 is represented by a transition. It does not matter whether the transition is
rising or falling, or whether the new level reached is high, 0, or low. Instead of a repeat-
ing sequence of the same binary value in the third waveform example, there is an alter-
nating binary sequence. Again, this pattern helps demonstrate that the absence of a
transition indicates a binary 0, and the presence of a transition indicates a binary 1.
Rising or falling edges indicate 1s. The very steep signal changes from one extreme to
the other, with a slight decrease in gradient at 0, indicate consecutive 1s. Any notice-
able horizontal line in the signal indicates a 0, or consecutive 0s.
Figure 6-12 is an example of 100BASE-TX signal taken from an oscilloscope. (In Fig-
ure 6-12, the y-axis is voltage; the x-axis is time. Voltage is measured as a differential

signal between 2.)
Figure 6-12 100BASE-TX Signal Sample
During the initial process of establishing synchronization with the link partner to
establish a link, the receiver circuit expects to see only 4B/5B idle code groups.
The cable pinout for a 100BASE-TX connection is identical to the one for 10BASE-T.
Two separate transmit/receive paths exist. For the connection between two stations or
two switches, a crossover cable is required. As for the connection between stations and
repeaters/multiport repeaters (hubs), a straight-through cable is used. Note that inside
an exclusively 100-Mbps hub is a bus topology, which is a collision domain. However,
if the hub is a 10/100 autosensing hub, which is vastly more common, the internal
topology is more complex, to account for the speed differential between 10BASE-T
and 100BASE-TX. It should be noted that, even though a 100-Mbps Ethernet hub is
significantly faster than a 10-Mbps hub, collisions are still a problem because both are
based on a shared bus architecture. Only through the use of switches and full duplex
are collisions avoided.
For the connection between a station and a switch, a straight-through cable is used.
The switch fabric circuitry allows full bandwidth simultaneously on multiple ports
without collisions.
1102.book Page 331 Tuesday, May 20, 2003 2:53 PM
332 Chapter 6: Ethernet Technologies and Ethernet Switching
Station-to-station, switch-to-switch, and station-to-switch connections in Fast Ether-
net all are point-to-point links: They have two physically separate communication
pathways/channels. In this case, collisions are not physical events, but rather the result
of an administrative decision to not allow simultaneous Tx and Rx. Therefore, 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.
However, station-to-hub connections must account for the bus topology within the
hub, a collision domain. Hence, this connection can only run half duplex and is subject
to CSMA/CD because of the physical nature of the structure.

Can a 100-Mbps technology allow 200 Mbps of traffic? 100BASE-TX carries 100 Mbps
of traffic in half-duplex mode (although some of this is overhead, not user data). But
100BASE-TX in full-duplex mode can exchange 200 Mbps of traffic (although, again,
some of this is overhead, not user data). The concept of full duplex will become increas-
ingly important with the desire to increase the speed of Ethernet links. You will learn
about the 100BASE-TX architecture rules later in this chapter.
100BASE-FX
Why use 100BASE-FX (introduced as part of the 802.3u-1995 standard)? At the time
copper-based Fast Ethernet was introduced, a fiber version was desired for backbone
applications, connections between floors and buildings where copper is less desirable,
and high-noise environments. 100BASE-FX also was positioned as an alternative to
the then-popular FDDI (100-Mbps dual fiber-optic Token Ring). However, the vast
majority of Fast Ethernet installations today are 100BASE-TX. One reason for the
relative lack of adoption of 100BASE-FX was the rapidity of the introduction of Giga-
bit Ethernet copper and fiber standards, which are now the dominant technology for
backbone installations, high-speed cross-connects, and general infrastructure needs.
The timing, frame format, and transmission were described previously in Chapter 5 and
are common to both versions of 100-Mbps Fast Ethernet considered in this chapter.
100BASE-FX uses 4B/5B encoded data with NRZI line encoding. Signals are LED pulses
on multimode optical fiber. NRZI encoding relies on the presence or absence of a tran-
sition in the middle of the timing window to determine the binary value for that bit
period. 100BASE-FX is synchronous.
Figure 6-13 illustrates the NRZI encoding examples. (The y-axis is optical power; the
x-axis is time.)
1102.book Page 332 Tuesday, May 20, 2003 2:53 PM
10-Mbps and 100-Mbps Ethernet 333
Figure 6-13 NRZI Encoding Examples
In the encoding examples in Figure 6-13, one timing window is highlighted vertically
through all four waveform examples. The top waveform has no transition in the center
of the timing window, so it is interpreted as a binary 0. No transition indicates that a

binary 0 is present. If the example waveform was all 0s on that line, the signal level
represented would be either low or high across the waveform. A single 1 was intro-
duced to indicate that 0s could be either level.
The second waveform has a transition in the center of the timing window. A binary 1
is represented by a transition. It does not matter whether the transition is rising or fall-
ing. Instead of a repeating sequence of the same binary value in the third waveform,
there is an alternating binary sequence. In this example, it is more obvious that no
transition indicates a binary 0, and the presence of a transition indicates a binary 1.
The NRZI-encoded, serialized bit stream is ready for transmission using pulsed light.
Because of cycle time problems related to turning the transmitter completely on and
off each time, the light is pulsed using low and high power. A logic 0 is represented by
low power, and a logic 1 is represented by high power.
Table 6-5 summarizes a 100BASE-FX link and the pinouts. A fiber pair with either
(ST) or (SC) connectors most commonly is used.
Table 6-5 100BASE-FX Pinout
Fiber Signal
1 Tx (LED and laser transmitters)
2 Rx (high-speed photodiode detectors)
1102.book Page 333 Tuesday, May 20, 2003 2:53 PM
334 Chapter 6: Ethernet Technologies and Ethernet Switching
Figure 6-14 shows an interface-to-interface fiber link. The two separate strands of mul-
timode fiber are often in the same cable structure, with dual connectors on each end.
Figure 6-14 Fiber Interface-to-Interface Connection
The MAC method treats the link as point-to-point, and fiber is intrinsically full duplex
because of separate Tx and Rx fibers. 100BASE-TX could run in half duplex, but this
would have length implications (the timing is actually a codeterminant, along with
attenuation/dispersion/fiber properties of length limitations and restrictions on numbers
of segments). Again, physical collisions between voltages are not an issue—these are
serial streams of light pulses on an optical fiber—but administratively can impose
CSMA/CD based on not allowing simultaneous Tx and Rx.

Can 100-Mbps fiber technology allow 200 Mbps of traffic? Somewhat analogous to
the separate transmit and receive paths in UTP, there are two such paths in 100BASE-
FX optical fiber, and 200 Mbps is possible.
Fast Ethernet Architecture
Fast Ethernet 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.
They are subject to the 100m UTP media distance limitation, but they have no limitation
on daisy-chaining.
Repeaters must be labeled with the word Class followed by a Roman numeral I or II
inside a circle, indicating Class I or Class II. A Class I repeater can introduce up to
140 bit-times of latency (delay). Any repeater that changes between one Ethernet imple-
mentation and another (for example, 100BASE-TX and 100BASE-FX) is a Class I
repeater. Also assume that any unlabeled repeater is a Class I device. Figure 6-15 illustrates
1102.book Page 334 Tuesday, May 20, 2003 2:53 PM
10-Mbps and 100-Mbps Ethernet 335
the maximum collision domain diameter for a Class I repeater for 100BASE-TX. Using
switches removes these restrictions, and the limiting factor becomes the media-determined
maximum length between interfaces.
Figure 6-15 Maximum Collision Domain Diameter for a Class I Repeater
A Class II repeater can introduce only a maximum of 92 bit-times of latency. Because
of the reduced latency, it is possible to have two Class II repeaters in a series, but only
if the cable between them is very short. Figure 6-16 illustrates the maximum collision
domain diameter for a Class II repeater for 100BASE-TX. Using switches removes
these restrictions, and the limiting factor becomes the media-determined maximum
length between interfaces.
Figure 6-16 Collision Domain Diameter for a Class II Repeater
As with 10-Mbps versions, it is possible to modify some of the architecture rules for
100-Mbps versions; however, there is virtually no allowance for additional delay. If
your network is implemented using new high-performance hardware, it is possible that

some of these limits can be exceeded. For example, if a longer cable is used between
repeaters, shorter cables would have to be used to each station. Modification of the
architecture rules strongly is discouraged for 100BASE-TX. Refer to the technical tim-
ing descriptions detailed in Clause 29 of the current 802.3 standard and the technical
information about your hardware performance before attempting it. Any device that
1102.book Page 335 Tuesday, May 20, 2003 2:53 PM
336 Chapter 6: Ethernet Technologies and Ethernet Switching
adapts between different Ethernet speeds, such as between 10 Mbps and 100 Mbps, is
an OSI Layer 2 bridge. It is not possible to adapt between speeds and still be a repeater.
However, the same device can repeat between ports linked at the same speed.
A 100BASE-TX UTP cable is about the same as a 10BASE-T cable, except that link
performance must meet the higher-quality Category 5 or ISO Class D requirements.
The 100BASE-TX cable between Class II repeaters cannot exceed 5m.
Links operating in full duplex can be substantially longer than what is shown in Table 6-6
because they are limited only by the capability of the medium to deliver a robust enough
signal for proper decoding, not the round-trip delay. It is not uncommon to find Fast
Ethernet operating in half duplex. However, half duplex is undesirable because the sig-
naling scheme is inherently full duplex, and forcing half-duplex communications rules
onto a full-duplex signaling system is not a wise use of resources.
It is recommended that all links between a station and a hub or switch be configured
for autonegotiation, to permit the highest common performance configuration to be
established with risking misconfiguration of the link. Disable autonegotiation and
force connection configurations only if autonegotiation fails or on certain selected con-
nections. The average station connection should be established by autonegotiation.
Table 6-6 summarizes the architectural rules for Fast Ethernet.
Table 6-6 Architecture Configuration Cable Distances
Architecture 100BASE-TX 100BASE-FX
100BASE-FX and
100BASE-TX
Station to station, station

to switch, switch to switch
(half or full duplex)
100m 412m —
One Class I repeater
(half duplex)
200m 272m 100m TX
160.8m FX
One Class II repeater
(half duplex)
200m 320m 100m TX
208m FX
Two Class II repeaters
(half duplex)
205m 228m 105m TX
211.2m FX
1102.book Page 336 Tuesday, May 20, 2003 2:53 PM
Gigabit, 10-Gb, and Future Ethernet 337
100BASE-TX 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, subject to the restrictions in Table 6-6 because of timing
considerations. The widespread introduction of switches has made this distance limita-
tion 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 Fast 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 logical topology and data flow, not timing or distance
limitations.
Gigabit, 10-Gb, and Future Ethernet
Fast Ethernet (100 Mbps) represented a major improvement over legacy Ethernet
(10 Mbps). Yet the even more rapid progression from Fast Ethernet to Gigabit Ether-

net is testimony to the power of IEEE standards, engineering advances, and market
forces. Gigabit Ethernet, 1000 Mbps, is a hundredfold increase in network speed
over the wildly popular 10BASE-T. Although MAC addressing, CSMA/CD, and,
most important, the frame format from earlier versions of Ethernet are preserved,
many other aspects of the MAC sublayer, the physical layer, and the medium have
been changed.
Copper interfaces capable of 10/100/1000 operation are now common. Gigabit switch
and router ports and blades are becoming routine in wiring closets. More multimode
and single-mode optical fiber is being installed. One major emphasis of Gigabit Ether-
net is optical fiber technology, but the need for a copper version—to use existing cable
plants and to use the ruggedness of copper in user environments—led to a very clever
scheme to get 1000 Mbps down the same Category 5 UTP used so successfully in
10-Mbps and 100-Mbps Ethernet. All of the Gigabit technologies are intrinsically
full duplex. The inexorable forward march of technology continues as standards and
technologies for 40 Gbps, 100 Gbps, and 160 Gbps currently are being implemented.
Most dramatic is the evolution of Ethernet from LAN applications only to an end-to-
end LAN, MAN, and WAN technology.
1000-Mbps Versions of Ethernet (Gigabit)
In 1998, the IEEE 802.3z committee adopted 1000BASE-X standard. This standard
raised the data transmission rate to 1 Gbps full duplex over optical fiber, a hundred-
fold increase in speed over 10BASE-T. The 1000BASE-T standard, specifying 1 Gbps
full duplex over Category 5 or higher UTP, was adopted in 1999.
1102.book Page 337 Tuesday, May 20, 2003 2:53 PM
338 Chapter 6: Ethernet Technologies and Ethernet Switching
Table 6-7 shows the parameters for 1000-Mbps Ethernet operation.
1000BASE-T, 1000BASE-SX, and 1000BASE-LX all share the same timing parameters.
Note that bit-time at 1000 Mbps = 1 nsec = .001 microseconds = 1 billionth of a sec-
ond. You also need to note that some differences in timing relative to legacy and Fast
Ethernet now are appearing because of the special issues that arise with such short bit
and slot times.

The 1000-Mbps (Gigabit) Ethernet frame has the same format as is used for 10- and
100-Mbps Ethernet. 1000-Mbps Ethernet has different paths for the process of con-
verting frames to bits on the cable, depending on which implementation is used.
Gigabit Ethernet is a tenfold increase in speed over Fast Ethernet. Just as with Fast
Ethernet, with this increase in speed comes extra requirements—the bits being sent get
shorter in duration (1 nanosecond), occur more frequently, and require more careful
timing. Their transmission also requires frequencies closer to medium bandwidth limi-
tations, and they become more susceptible to noise. In response to these issues of syn-
chronization, bandwidth, and signal-to-noise ratio, Gigabit Ethernet uses two separate
encoding steps. The basic idea is to use codes—which can be engineered to have desir-
able properties—to represent the user data in a way that is efficient to transmit, includ-
ing synchronization, efficient usage of bandwidth, and improved SNR characteristics.
Table 6-7 Parameters for Gigabit Ethernet Operation
Parameter Value
Bit-time 1 nsec
Slot time 4096 bit-times
Interframe spacing 96 bits*
*The value listed is the official interframe spacing.
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)
Burst limit 65,536 bits
1102.book Page 338 Tuesday, May 20, 2003 2:53 PM