Sei sulla pagina 1di 31

CpE

p 490 Information System


y
Engineering I

Prof. Hong Man

Department of Electrical and


Computer Engineering
Stevens Institute of Technology
Chapter 2

Underlying
d l i Networkk Technologies
h l i
Layered Network Structure
Two Approaches to Networking

• Connection-oriented ((circuit-switched)) networks:


– Establishing a dedicated connection or circuit between
the sender and the receiver.
– Example: telephone system, N-ISDN. Dedicated data
path with ~64 Kbps bandwidth.
– Advantage: traffic is relatively predictable,
predictable medium
access control resource reservation and quality of
service (QoS) guarantee are possible.
– Disadvantage: require a call set-up delay, high cost,
possible waste of bandwidth.
Two Approaches to Networking

• Connectionless (p(packet-switched)) networks:


– Divide data into small pieces (packets), e.g. a few
hundred bytes.
– Each packet carries identification information for
network hardware (routers, switches) to pass the packet
to its destination.
– Advantages:
Ad
• multiple conversations can be carried at the same
time – low cost, bandwidth efficiency;
• no call set-up procedure, transmission at any
moment.
– Disadvantage: no capacity guarantee,
guarantee large delay
variation.
WAN and LAN

• Packet switchingg technologies


g are divided into wide area
networks (WANs) and local area networks (LANs).
– WAN technologies provide communication over long
distances. Lower speed (1.5 Mbps ~ 155 Mbps), longer
delay (0.001 ~ 0.5 sec.).
• WAN has complex configurations, relies on switches
(or routers).
routers) The larger the WAN size becomes,
becomes the
more the switches are.
– LAN technologies span small areas. Higher speed (10
Mb ~ 2 Gb
Mbps Gbps),) shorter
h ddelay
l (0(0.0001
0001 ~ 0.01
0 01 sec.).
)
• LAN has simple network configurations, relies on
p
end computer-network interface devices – network
interface card (PCMCIA card, PCI board etc.)
WAN and LAN

• Hardware addresses: each network hardware system


y
defines its own addressing mechanism to specify the
destination of a packet. Each computing device connected
to the network has a unique address.
address A sender must know
the address of the receiver.
Conventional Shared-Medium LANs

• In a conventional local area network ((LAN),) a single


g
medium (usually a cable) is used.
• All the end-systems (or stations) are sharing this common
medium. Each station can hold several simultaneous
communications through its connection to the medium.
• Transmissions are baseband, therefore the multiplexing is
achieved through time division – time division multplexing.
multplexing
• A data stream is broken into a series of frames, which are
transmitted during the time slices allocated temporary to
the
h station
i based
b d on certain i medium
di access mechanism.
h i
• Medium access control (MAC) is a sub-layer of data link
y in the layered
layer y pprotocol structure.
LAN Topologies
Ethernet

• Ethernet was developedp in the 70’s byy Xerox,, Intel,, DEC.


A standard was released by IEEE (802.3). Ethernet
becomes the dominant LAN technology.
• Ethernet
Eth t cabling:
bli
– 10Base-5 coaxial cable: a thick 50 ohm (yellow) cable,
maximum
a u length
e gt of
o 500 m per
pe segment,
seg e t, with
w t 100
00
machines (nodes) per segment .
– 10Base-2 coaxial cable: a thinner 50 ohm cable,
maximum
i length
l h off 200 m per segment, with
i h 30 nodes
d
per segment.
– 10Base
10Base-T T twisted pair: 100 m per segment, 1024 nodes.
Ethernet Capacity

• 10 Mbpsp ((standardized in the 70’s)) uses 10Base cables.


• 100 Mbps fast Ethernet, 100Base-T technology can use the
same cable used in 10Based-T. The design was not
optimized
ti i d for
f increasing
i i highest
hi h t possible
ibl throughput
th h t
between two end systems, instead it was for increasing
number of stations and aggregated traffic.
• 10/100 dual-speed Ethernet, cabling and connectors used
in 10Base-T and 100Base-T are compatible. The hardware
can automatically detect the transmission speed of a
particular system, because 100Based-T hardware interjects
extra signals.
Ethernet Capacity

• 1000 Mbps p ((Gigabit)


g ) Ethernet developed
p in the late 90’s
for corporate backbone. Need different wiring. Designed
for increasing total traffic.
• All diff
differentt data
d t rate
t Ethernet
Eth t use the
th same packet
k t format
f t
and maximum packet size.
Ethernet Properties

• Bus topology.
p gy
• Broadcast (multicast) technology.
• Each host interface device determines the packet
destination.
• Best-effort delivery, hardware provides no information
about successfulness of packet delivery.
delivery (Higher level
protocol can do this job.)
• Ethernet medium access control (MAC) protocol: carrier
sense multiple access with collision detection
(CSMA/CD).
CSMA/CD

• Anyy data transmission is ppropagated


p g in both directions on
the medium.
• Before transmission, a station first listens to the channel. If
th channel
the h l is
i idle,
idl the
th station
t ti starts
t t transmission,
t i i
• If the channel is busy, the station will wait:
– until the channel becomes idle and then starts to transmit data
(1-persistent), OR
– for a random amount of time and listen to the channel again
(nonpersistent), OR
– until the channel becomes idle and then start transmission with a
probability of p, or wait till the next time slot with a probability of
1-p
1 p (p
(p-persistent).
persistent).
CSMA/CD

• A collision can be detected byy anyy station if it sense an


abnormal electrical level.
• If a collision occurs, all stations halt their transmissions
andd wait
it a random
d amountt off time
ti (with
( ith binary
bi
exponential backoff policy) and starts all over again.
Ethernet Hardware Address

• Ethernet uses 48-bit addressingg scheme.


• Addresses are assigned to all Ethernet interface hardware
by manufactures.
• Change an Ethernet interface hardware will change the
Ethernet address of the machine.
• Ethernet addresses are also called physical addresses,
addresses or
media access addresses, or level-2 addresses.
• Interface ignores all the packets that are addressed to other
machines, and only pass to the host the packets addressed
to it.
Ethernet Hardware Address

• Ethernet address can be:


– a unicast address, addressing to one interface,
– a multicast address, addressing to a group of interfaces.
A multicast group member (host) should inform its
network interface device to accept the multicast
add ess.
address.
– a broadcast address (all 1’s), addressing to all stations
on the Ethernet.
Ethernet Frame

• Ethernet is a link-level ((MAC)) connection amongg


machines, so its data packets are usually called frames.
• Ethernet frame: ≥64 octets (bytes) and ≤1518 octets,
i l di all
including ll header,
h d datad t andd CRC
• Figure 2.2
Ethernet Frame Format

• Preamble: 64 bits of alternatingg 0’s and 1’s for


synchronization.
• Destination Ethernet address
• Source Ethernet address
• Frame type: 16-bit integer to identify the type of the frame,
specifying the protocol to be used.
used
• Data
• CRC: 32-bit Cyclic
y Redundancy y Check,, detecting
g
transmission errors.
Extending an Ethernet

• To extend an Ethernet beyond


y the maximum length
g of the
cable with repeaters
– Repeater: an electronic device
operators on analog signals, and
relays all electronic signals from
one cable
bl to
t another.
th
– In order to preserve the CSMA/CD
timing at most two repeaters can
timing,
be placed between two machines.
Extending an Ethernet

• Connectingg two Ethernet with bridge


g
– Bridge: operates on packets instead of electronic
signals.
– It validates frames before accept them and transmit
them to another Ethernet segment. So it does not repeat
noise,
o se, eerror
o or
o malformed
a o ed frames.
a es.
Extending an Ethernet

• Connectingg two Ethernet with bridge


g ((conti.))
– Bridge follows the CSMA/CD rules, so collisions and
propagation delays on one segment remain isolated
from those on the other.
– An arbitrary number of Ethernet segments can be
connected by bridges.
– An
A adaptive
d i bbridge
id can learn
l which
hi h hosts
h lie
li on which
hi h
Ethernet segment form frame addresses, therefore it
needs no configuration.
• Repeaters and bridges are transparent to host machines,
high level protocols do not have to consider their
existences.
Token Rings

• Two types
yp of token ringsg operate
p at 4 Mbps
p and 16 Mbps.p
The achievable end-to-end bit rate between two stations
may reach 3.8 Mbps and 15.5 Mbps respectively.
Token Rings

• Token Rings g use the token ppassingg pprotocol ((IEEE 802.5):


)
– The Token Ring uses the ring topology.
– Data transmissions are unidirectional.
– The data frames circulated on the ring, from station to
station. A small delay will occur when a frame passes a
station (1 ~ 8 bits).
bits)
– A particular bit pattern (token) is circulates on the ring
at full speed when the ring is idle.
Token Rings

– A token mayy have two statuses:


• Free: no significant data appended to it,
• Busy: it is followed by data.
– When a station want to transmit data, it first waits for
the token to arrive
• If the
th token
t k isi busy,
b it will
ill keep
k waiting
iti
• If the token is free, it will
– Set the token status to busy,
busy
– Append the data frames,
– Specify
p y the destination in the frame header.
Token Rings

– When a station sees a busyy token, it will check the


destination address of the data frame.
• If it belongs to other station, it will pass on.
• If it belongs to the current station,
station it will
– Copy the data frame to its buffer,
– Set a control bit to signal “data copied”,
– Forward the modified data frame.
– When the data returns to the sender station, the sender
will remove the data frame from the ring and set the
token to “free”.
• Token Rings cabling: shielded twisted pairs.
Fiber Distributed Data Interconnect

• Fiber Distributed Data Interconnect ((FDDI)) is a high


g speed
p
LAN technology
– 100 Mbps, the achievable end-to-end bit rate is usually
aroundd 50 ~ 60 Mbps,
Mb sometime ti even up to t 90 Mbps.
Mb
• At these high transmission speeds, the actual
tthroughput
oug put highly
g y depends
depe ds on
o the
t e performances
pe o a ces anda d
processing power of the participating stations and
high level protocols.
– Optical
O i l fiber
fib network:k high
hi h speedd andd low
l interference
i f
(low noise).
Fiber Distributed Data Interconnect

• FDDI cabling:g
– Multimode optical fiber (MMF): uses a light-emitting
diode (LED) as the source of the optical signal at a
wavelength
l th off 1325 nm.
– Single-mode optical fiber (SMF): uses a leaser diode
ttransmitter
a s tte as the
t e source.
sou ce.
– Low-cost optical fiber (LCF): may use cheap CD lasers
at a wavelength of 1300nm.
– Unshielded twisted pair (UTP): various qualities.
FDDI Properties

• Ringg topology
p gy
• Token passing ring protocol (ANSI X3T9.5)
• Distinct features in FDDI rings:
– T
To bbe able
bl tto transmit,
t it a station
t ti mustt capture
t the
th token.
t k When
Wh
transmission is completed, the station will issue a new token into
the ring.
– Two frames can be transmitted at the same time in two different
sections of the same ring.
– A station can not keep the token longer than the token holding time
(THT), during which the station can transmit several frames.
– A target token rotation time (TTRT) is set for all stations to ensure
that the token rotates fast enough such that each station will get
access at least once within the TTRT.
FDDI Properties

• Both Token Ringg and FDDI ring g have a second ringg as


backup in case of ring breakage.
– The direction of data circulation in the backup ring is opposite to
the direction in the first ring.
ring
– Self healing: when an node failure occurs, a neighboring FDDI
interface can detect this hardware error, and bypass the failed node
using the backup ring.
ring
FDDI Frame Format

• Maximum frame size 9000 symbols


y = 4500 octets.

Potrebbero piacerti anche