evolution of ethernet efficiency
TRANSCRIPT
EVOLUTION OF ETHERNET EFFICIENCY
Gurkamal Deep Singh Rakhra
Student at University of Texas at Arlington
Vidisha Shah
Student at University of Texas at Arlington
Yash Gupta
Student at University of Texas at Arlington
Shrey Khosla
Student at University of Texas at Arlington
Yash Shah
Student at University of Texas at Arlington
Through the years Ethernet technology has evolved to keep pace with ever increasing demands for higher bandwidth. In
its early days, Ethernet bandwidth was limited to 1-2 Mbps, and today we see installations supporting 10 Gbps as
common. As the bandwidth has increased, Ethernet designers have been challenged to maintain high utilization
efficiency of the available network capacity, often requiring changes in the structure and/or underlying technology of
Ethernet LANs. We show how Ethernet evolved during the times and what changes occurred because of which such high
speeds are achievable now-a-days.
1. INTRODUCTION
Ethernet is an environment which consists of a
network connecting a number of computer or similar
devices with set of protocols to monitor the exchange
of information between similar or different devices.
This environment forms a Local Area Network
(LAN). Ethernet has evolved to a great extent in
recent years to meet the need of current speeds in the
computer industry. Starting from 3 Mbps we have
now managed to attain speed of 100 Petabytes per
second. This has been possible because of
enhancements in old technology and development of
new technology. Started with the use of copper wire
using electric signals to send just voice information,
now we can send voice and data simultaneously at
very high speeds using light signals.
2. MOTIVATION
Bob Metcalfe was the person who invented Ethernet
based on an earlier experiment called Aloha
Network. In Aloha protocol, the Aloha station could
send anytime it liked and waited for an
acknowledgement. If acknowledgment was not
received within a certain period of time, the station
assumed that another station had transmitted at the
same time causing a collision. On detecting collision,
both stations set a random back off time and
retransmit their packets with a better probability of
success. The problem with this system (Pure Aloha)
was that maximum channel utilization was about 18
percent due to increasing rate of collisions with
increasing load. Soon an upgrade of this system was
developed which was called Slotted Aloha which
assigned transmission slots and used a master clock
to sync transmissions, which resulted in increased
utilization of the channel to 37 percent.
Metcalfe thought that he could improve the Aloha
System of random access to a shared communications
channel. He developed a system that detected
collisions, sense the carrier before transmitting and in
which multiple stations could access the channel
which gave the name of Carrier Sense Multiple
Access with Collison Detection (CSMA/CD) to
Ethernet. Metcalfe developed a much more complex
and efficient backoff algorithm which in combination
with the CSMA/CD gave the Ethernet system the
channel utilization of up to 100 percent.
Soon with his experiments, Metcalfe and his
colleagues developed first experimental Ethernet
system to interconnect Xerox Alto. This experimental
network was named the Alto Aloha Network. In
1973, the name was changed to “Ethernet” by
Metcalfe to convey that this system could support
any computer.
Figure 2.1 Diagram of original Ethernet
System by Bob Metcalfe in 1976
3. EVOLUTION OF SPEED
Ethernet initially supported 10 Mbps speed which
was sufficient for the then available computers. The
first Ethernet standard was IEEE 802.3 standard for
thick Ethernet also known as 10BASE5. The problem
with thick Ethernet was that it was very difficult to
install a thick coax cable in a building and connecting
computers to the cables was a big challenge. After
that thin coaxial Ethernet was standardized by IEEE
802.3 committee in mid-1980’s. It was still difficult
to manage Ethernet systems based on coaxial cables
as they are based on bus topology, in which every
computer sends signals over a single bus. If the cable
fails, the whole network came to a halt and
troubleshooting the cable can take a long time.This
development also included the twisted-pair and fiber
optic varieties for the 10 Mbps system. This
development was based on more reliable star
topology, in which the computers are networked to a
central hub. This arrangement was much easier to
install and manage, also troubleshooting was easier
and quicker as well. Soon 100 Mbps Fast Ethernet
system was developed which included varieties of
twisted-pair and fiber optic media systems. The latest
development is called the Gigabit Ethernet system
which was developed using both fiber optic and
twisted-pair cabling. The Ethernet standard defines
three fields:
a. Data rate
b. Modulation type
c. Additional distinction
For example:
10BASE5
In this example, 10 means 10 megabits per second
transmission speed, BASE means baseband
transmission and 5 refers to 500 meter maximum
segment length. Similarly we have 10BASE2,
10BROAD36, 1BASE5 etc. Later, instead of the
distance, type of cable was written. For example:
10BASE-T
10 and BASE have the same meaning, but T stands
for Twisted-pair. Similarly F stands for Fiber Optic
cable. 100BASE-T, 10BASE-F are few examples.
4. EVOLUTION OF MEDIA
Media can be categorized in two ways- Wired and
Wireless media. Wired media consists of all the
different types of cables like fiber optic, twisted-pair
etc. And wireless covers Wireless Fidelity (Wi-Fi).
Ethernet initially consisted of just wired media
(coaxial) and now-a-days we have fiber optics.
Figure 4.1 Coaxial Cable
Co-Axial cable: Co-axial cable is used for data
communications. There is a single centre solid wire
covered by insulating dielectric wrapped by foil
conductor. All these solid wire, insulating dielectric
and foil conductor is covered by outside insulation.
This insulation is responsible for cable’s inductance
and capacitance. There is one wire that carries signal
surrounded by another concentric physical channel
running on same axis, that’s why it is called
“coaxial”.Depending upon the technology used
different kind of cables are used. All cables are
identified according to its impedance or RG-type.
Coaxial cable is typically designed as 50
ohm, 52 ohm, 75 ohm, and 93 ohm
depending upon the application
Power loss depends on the cable design and
is both frequency and length dependent
Losses are based on signals reflecting back
to sources rather than propagating through
cable.
Coaxial cable supports 10 to 100 Mbps and is
relatively inexpensive.
Twisted-Pair cable: Twisted pair cables were
invented in the year 1881 by Alexander Graham Bell
[1]. The concept of twisted wire came into existence
to solve the problem of noise and cross talk that
existed in open wire that were earlier used. The
twisted pair cable has 2 insulated wires, a conductor
that is copper or an alloy (can be silver plated)
insulated by plastic which is arranged in a spiral
pattern as seen in the figure 3.2.1.The signal is
transmitted through one wire of the pair and a ground
reference is transmitted through the other wire. This
twisting of the wires helped in skewing the electro-
magnetic polarity of the noise or the unwanted signal.
This skewing effect is called canceling effect, and
more the twist of the wire per inch the better is the
effect. [2]
Figure 4.2 Twisted Pair Cable
Types of Twisted Pair Cables
Unshielded Twisted Pair (UTP) cable: Unshielded
Twisted Pair (UTP) is a type of twisted cables. UTP
has usually 4 pair of wires that is in all 8 wire in 1
cable and that is covered by an outer jacket, the
architecture for the same can be seen in figure 3.2.2.
UTP cables support speed and throughput of 10
Mbps to 100 Mbps. It does not cost much (least
expensive). The media and connector size for UTP
cable is small. The cable length at a stretch is just 100
meter there after it needs a repeater to maintain the
quality of the signal [1]. The most common UTP
connector is RJ45 (Registered Jack-45). In RJ- 45
there are 2 different types of wiring sequences, this
sequence is based on the color codes of the wires.
Figure 3.2.3 shows both the standards, i.e. T 568-A
and T 568-B. T 568-A is commonly used
sequence. The color scheme for T 568-A is
Pair 1 = White/Blue and Blue.
Pair 2 = White/Orange and Orange.
Pair 3 = White/Green and Green.
Pair 4 = White/Brown and Brown.
The arrangement of the pairs for both the types of
connector is seen in the figure 3.2.3. The position of
the pair 2 and 3 is reversed in the other connector [2].
Figure 4.3 Unshielded Twisted Pair Cable
Figure 4.4 RJ-45 Color Sequence
There are 6 categories (Cat 1, Cat2, Cat3, Cat4, Cat
5, Cat5e, and Cat6) of UTP based on the bandwidth
they support and their application. The list of this
categories their bandwidth and application is given
below in table 3.2.1.
Cat1:- It is a low quality cable just suitable for voice
transmission (Analog transmission). It supports
bandwidth of 0.4 MHz and data rate less than 100
Kbps. It is not use anymore mainly found in very old
buildings.
Cat2:- It is good for voice transmission and also
supports low data rate (analog as well as digital
transmission). It has bandwidth of 4 MHz and data
rate up to 4Mbps. It just supports low speed token
ring network.
Cat3:- It is an improvised version of Cat 2. Twist
Rate for Cat3 is a minimum of 3 twists every foot of
the cable. It is commonly found in residential
building for phone networks. It has a bandwidth of 16
MHz and supports data rate of 10 Mbps. It is used in
voice application and 10baseT Ethernet. It has an
attenuation of 11.5 dB and impedance of 100 ohms.
Cat4:- It is an improvised version of Cat 3 which is
usually used in Token Ring networks. It has
bandwidth of 20MHz and data rate of 16Mbps.It is
used in 10baseT Ethernet, has an attenuation of 20
MHz It has an attenuation of 7.5dB.
Cat5:- It is used for high speed data transmission. It
has a bandwidth 100 MHz of and data rate of 100
Mbps. It allows a maximum distance of 100 meters.
Used in 10Mbps and 100 Mbps Ethernet. It has an
attenuation of 24.0dB.
Cat5e:- It is an extension of Cat5 that includes an
extra feature that minimize the crosstalk effect. It has
a bandwidth of 100 MHz and data rate of 100 Mbps
and 1000 Mbps. It is used in Fast Ethernet and
Gigabit Ethernet. It has an attenuation of 24.0dB. It is
common for high speed data communication.
Cat6:- It has more twist rates as compared to that of
Cat5 that makes it better in reducing the interference
of noise or unwanted signal. It has a bandwidth of
250 MHz and a data rate of 2500Mbps. It is used in
Gigabit Ethernet and 10 Gig Ethernet (10000 Mbps).
Catego
ry
Band-
width
(MHz)
Data
rate
(Mbp
s)
Application
Cat 1 0.4 0.1 Telephone
communication
Cat 2 4 2 Voice and low data
network
Cat 3 16 10 Voice and data network
Cat 4 20 16 16- Mbps Fast Ethernet
Cat 5 100 100 100 Mbps Fast Ethernet
Cat5e 100 1000 1000 Mbps Gigabit
Ethernet
Cat 6 250 2500 Gigabit Ethernet and 10
Gig Ethernet
Table 4.1.1 Categories of UTP
Shielded Twisted Pair (STP) cable: Shielded
Twisted Pair (STP) is a type of twisted cable where
the various pair of wires are shielded. So a special
protection is provided to each pair thereafter there is
an overall shield and then there is an outer jacket.
The figure 3.2.4 shows the architecture of STP
cables. The extra shield over each pair helps the
problem of cross talk (i.e. data UTP cables support
speed and throughput of 10 Mbps to 100 Mbps. It is
moderately costly (expensive then UTP because of
the extra shields). The media and connector size for
UTP cable is medium to large and the cable length at
a stretch is just 100 meter there after it needs a
repeater to maintain the quality of the signal. The
architecture of the STP connector is similar to UTP
connector [1]. As seen above UTP has 6 categories of
cable from Cat 1 to Cat 6. Similarly STP also has 2
Category Cat 7 and Cat 8. The Category is
differentiated on its bandwidth and data rate
supported.
Figure 4.5 Shielded Twisted Pair Cable
Optical fiber cable: The technology used in this
cable is transmitting data through light signals. These
cables support very high speeds and are very
expensive. The phenomena used here is total internal
reflection. Optical fiber consists of a core and
cladding, which has difference in their refractive
index resulting in total internal reflection. On the
sending side, the electric signals are converted in to
light signals which on the receiving end gets
converted in to electric signals again with the help of
transceiver. Optical fiber is now used in almost every
telecommunication companies to transmit internet
communication, telephone signals and cable
television signals. Advantages of this cable is that it
is very efficient as compared to other cables in terms
of maintenance, speed and disturbances. [3]
Figure 4.6 Optical fiber cable
Following are the advantages of fiber optic cables:
Immunity to electromagnetic interference,
including nuclear electromagnetic
pulses (although fiber can be damaged
by alpha and beta radiation).
High electrical resistance, making it safe to
use near high-voltage equipment or between
areas with different earth potentials.
Lighter weight—important, for example, in
aircraft.
No sparks—important in flammable or
explosive gas environments.
Difficult to tap the signal without disrupting
the signal. It is an important feature in high-
security environments.
Much smaller cable size—important where
pathway is limited, such as networking an
existing building, where smaller channels
can be drilled and space can be saved in
existing cable ducts and trays.
Resistance to corrosion due to non-metallic
transmission medium.
Distance Bandwidth
Copper 2.5 km 1.5 Mb/s
Fiber 200 KM 2.5+ Gb/s
Table 4.1.2 comparison of copper and
fiber optic cable [4]
5. EVOLUTION OF DEVICES
Ethernet hubs: Initially it was capable of supporting
only a theoretical maximum of 10 megabits per
second data. The maximum theoretical rate for an
Ethernet hub increased to a high amount in a very
less time. In a short span of time the maximum rate
was raised to 100 megabits per second. The fastest
Ethernet hubs now supports a maximum rate of 1000
megabits per second. The work of a LAN is to
connect many more devices than just two systems, so
that a simple cable two-ways doesn't get the job done.
Fast Ethernet uses the same CD/CSMA as Ethernet
but because of a limitation in the usage of repeaters
and cable length restrictions makes it difficult to
detect the collisions in a short span of time. At an
early stage hubs supporting 10/100 Mbps started to
get used where 10Mbps were connected to other
same speed systems and 100Mbs to the 100Mbps
systems. And hence technically this hubs acted as a
bridge between the devices.
After few years it was targeted to bridge all the ports
in the network to increase the connectivity level. And
then this multiport devices were named as Switching
hubs or Ethernet Switches. Switches are smarter as
they get to know about the addresses which are
reachable over a port by just an observation of the
source address in the packet flowing through it. The
difference between a switch and a hub is that a hubs
is used to broadcast the packets to all other links
whereas a switch is comparatively smarter and only
sends packets to links that require it. So switches will
always need to be powered to run their logic chips.
[8]
Ethernet Switches: Ethernet provides a central
connection in network giving each device its own
link and full bandwidth. Divides LAN data into
smaller segment and sends only those segments
required to reach PC’s. Each port on switch
represents a dedicated 10-100 Mbps pathway as each
user is allotted such bandwidth. [5]
Earlier Ethernet were switched with long runs of co-
axial cables attaching multiple stations which can
transmit data at 10Mbps whereas Nowadays Ethernet
Network uses twisted pair wiring or fiber optics to
connect stations which can operate at 100 or even
1000 Mbps. Switched network replace older version
of Ethernet with dedicated segment for each station
which connect to a switch acting like a bridge
connecting many single stations. [6]
Reasons to switch to Ethernet Switches
1) Technically more advanced and efficient
then earlier used hubs, repeaters.
2) Allows devices to work in full duplex mode.
3) It can easily recognize pieces of data along
with exact location they are intended to
place.
4) They can save and detect MAC address of
linked computers making flow of bandwidth
lighter.
Designed to be immediately ready for usage so
preferred by users, companies and used in wide
networked environments. [7]
6. WIRELESS MEDIA
802.11
Introduction: Wireless Ethernet is a technology that
enables two or more entities to communicate without
network cabling. It allows to use the device when
they are in the periphery of the wireless network and
use the service.[10]
History: Wireless Ethernet was first developed by
Norman Abramson in 1970s and it was the first
wireless communication network and it was
developed at University of Hawaii. After this, there
was huge development in the field of wireless
technology and hence in late 1990s it resulted in
making the IEEE 802.11 standards. Before there
exists many proprietary systems but they were very
costly hence they were not successful. In 1991 AT&T
and NCR invented a precursor to 802.11 and it was
initially designed to use for cashier system. The first
version of 802.11 was released in 1997 and it
provided a speed of 2Mbps.[10]
Advantages of Wireless Network [9]
1. Easy to install.
2. Easy to add more computers to the networks.
3. Promotes flexibility.
4. Supports mobility.
5. Works over great distances.
6. Configurations can be easily changed.
Disadvantages of Wireless Network [9]
1. Wireless network setup costs around times more
than that of wired network.
2. The base station or access points used in
infrastructure network cost more than that of hubs
and wires used in wired network.
3. It has slower bandwidth as compared to that of
wired.
4. It depends on environmental conditions.
5. Wireless network is more susceptible to packet
loss as compared to wired network.
802.11 Wireless LANs
1. 802.11a: It is an amendment made to the
original 802.11 standard in 1999. It uses the
same core protocol as the original 802.11. It
operates on 5 GHz band, and uses an
orthogonal frequency multiplexing (OFDM).
It has a maximum data rate of 54 Mbps. It
originally has 12/13 non-overlapping
channels. It avoids signal interference from
other users like cordless phones. The 2.4
GHz band is heavily used to the point of
being crowded and hence 5 GHz is having
the advantage that it is not crowded. The
conflicts caused in such a crowded
connection can cause frequent dropping of
packet and the service is degraded. The
disadvantage is that the signals that cannot
penetrate as far as those for 802.11b because
they are mostly absorbed by the walls and
other solid objects in their path.[11]
2. 802.11b: In July 1999, the 802.11a was
expanded and a new 802.11b specification was made
which supports a bandwidth of the 11 Mbps.
comparable to traditional Ethernet. It uses 2.4 GHz of
the unregulated radio signaling frequency which is
used in the original Ethernet standards. The 802.11b
gets inference from microwave ovens and other such
appliances which use the same frequency channel i.e.
2.5 GHz. The advantage of 802.11b standard is low
costs, and it provides good signal range and they are
not easily obstructed. The disadvantage is that it
having slow speed, and as it works on unregulated
frequency band it gets interference from home
appliances.[12]
3. 802.11g: In the advent of 2003 and 2004, a
newer standard of WLAN product which supports a
newer standards 802.11g came into the market. It
combined the best of both 802.11a and 802.11b.
802.11g standard supports 54 Mbps of bandwidth
like 802.11a and it works on 2.4 GHz frequency like
802.11b which has a greater range. 802.11g is
backward compatible with 802.b. Hence the access
points using the 802.11b can connect with 802.11g
devices. The advantage of this standard is that it
provides fast maximum speed, its signal range is
good and it is not easily obstructed. The
disadvantages of this standard is that it is costlier than
802.11b devices, and the home appliances may still
interfere on the unregulated signal frequency.[12]
4. 802.11n: In 2007 there was an amendment
made to the 802.11 which was termed as 802.11n. It
was designed to improve the bandwidth supported by
using multiple signals and antennas. It was again
amended in 2009 so that it can provide 300 Mbps of
network bandwidth. It provides improved signal
intensity than its predecessors. It is also backward
compatible with 802.11 b/g. The advantage of this
standard is that it is fastest maximum speed and its
gives the best signal range. It is more resistant to
interference in signals from the other signals. The
disadvantages are that it is not yet fully finalized. It is
more costly than 802.11g. The use of multiple signals
can cause interference from other devices which use
both 802.11b and 802.11g standards. [12]
5. 802.11ac: In 2008 there was an amendment
made to 802.11 standard which was termed as
802.11ac. Its development started in 2008 and
continued till 2013. It is utilizes a dual band wireless
technology, which supports 2.4 GHz and 5 GHz both
frequency. It provide bandwidth up to 1300 Mbps on
5GHz band and 450 Mbps on 2.4 GHz. It is also
backward compatible with 802.11 b/ 802.11 g/
802.11 n.[12]
802.11
a
802.11
b
802.11
g
802.11
n
802.11
ac
Frequen
cy
5-6
GHz
2.4-2.5
GHz
2.4-2.5
GHz
2.4-2.5
GHz
2.4 &
5 GHz
Data
Rate
upto
54
Mbps
upto
11
Mbps
upto
54
Mbps
upto
200
Mbps
upto
1300
Mbps
Modulati
on
OFD
M
DSSS OFD
M
SDM OFD
M
Table 6.1 Specifications
Types of Wireless Network
Infrastructure Mode:-
Base Station
Wireless hosts
Wireless links
Figure 6.1 Infrastructure network
In infrastructure mode all the devices are connected
to a single access point (base station) and they
communicate through access point. The packet is sent
to access point and access point future send it to the
destination.[13]
Ad-hoc mode:- Ad-hoc network(Peer-to-Peer) does
not have a centralized base station/access point to
communicate with other devices. The devices are
directly connected to each other with a wireless link
along the coverage area. [13]
Wireless hosts
Wireless links
Figure 6.2 Ad-hoc mode
Various Parameters Used for Analysis of Wireless
Networks
Throughput: Throughput is rate at which messages
are delivered successfully over the network. It is
measured in bits/second.
Packet Delivery Ratio: It is the ratio at which the
number of packets are received to number of packets
generated.
Packet Loss: When one or more packets fail to
reach the destination across the network then it is
called packet loss.
Bit Error Rate/Bit Error Ratio (BER): It is the
number of bits altered because of noise dictation and
interference to the total number of bits transferred in
a particular time interval. It is also defined in terms of
Probability of Error (POE).
Table 6.2 Comparison of various parameters with
respect to number of packets send.[14]
7. CONCLUSION:
In this paper we have seen how Ethernet has emerged
from use of copper cables to use of fiber cables. We
also saw the evolution in speed starting from 10
Mbps to 100 Gbps. Different environments need
different types of cables to be installed. With the
change in computer systems, the Ethernet
architecture also changed to support higher speeds.
REFERENCES
[1]http://en.wikipedia.org/wiki/Twisted_pair
[2]http://www.infocellar.com/networks/cables
/twisted-pair-cables.htm
[3]http://en.wikipedia.org/wiki/Fiber-
optic_communication
[4] http://www.thefoa.org/tech/fo-or-cu.htm
[5]http://www.blackbox.com/resources/blackboxexpl
ains.aspx?id=bbe_4170
[6]http://computer.howstuffworks.com/ethernet14.ht
m
[7]http://ezinearticles.com/?Ethernet-Switching---
What-You-Should-Know&id=4696844
[8]http://arstechnica.com/gadgets/2011/07/ethernet-
how-does-it-work/2/
[9]http://en.wikipedia.org/wiki/Wireless_LAN
[10]http://en.wikipedia.org/wiki/IEEE_802.11a-1999
[11]Computer Networking: A Top Down Approach -
6th edition Jim Kurose, Keith Ross Addison-Wesley
[12]http://airccse.org/journal/ijasuc/papers/0910ijasu
c03.pdf
[13]http://compnetworking.about.com/cs/wireless802
11/a/aa80211standard.htm
