Transcript Chapter 1

INFO 330
Computer Networking
Technology I
Chapter 1
Networking Overview
Jennifer Booker
INFO 330
Chapter 1
1
www.ischool.drexel.edu
Computer Networks
• A network is the structure that allows
computer applications to communicate
with each other
– The applications could be executed by the
user, or part of the operating system
• Not every computer system is designed to
allow networking
– Microsoft DOS had no native networking
ability; it was added after the need arose
INFO 330
Chapter 1
2
www.ischool.drexel.edu
The Internet
• The Internet is the primary model for
understanding networking concepts
because, well, nearly every computer and
many other things could be connected to it
INFO 330
Chapter 1
3
www.ischool.drexel.edu
The Internet
• Key parts of any network include
– Hosts or end systems, which are the
computers and other things with which most
people interact
• End user computers, workstations, and servers are
all considered hosts
• As of July 2008 there were about 600 million hosts
on the Internet, and about 850 million as of July
2011
INFO 330
Chapter 1
4
www.ischool.drexel.edu
The Internet
– Communication links,
which are the wired or
wireless means used to
connect to the network
– Packet switches, which
help guide information
between hosts
• Routers and link-layer
switches are the
primary types of
packet switches
Graphics are taken from the text’s lecture notes
INFO 330
Chapter 1
5
www.ischool.drexel.edu
The Internet
• The network sends chunks of information
called packets along a route or path to get
from one host to another
– The speed at which it does so is the
transmission rate, typically in bits per second
(bps)
INFO 330
Chapter 1
6
www.ischool.drexel.edu
The Internet
• The control over choosing the path is
known as packet switching
• End systems connect to the Internet
through an Internet Service Provider (ISP)
• ISPs provide many levels of service
– Residential or business service, typically from
56kb dialup to DSL, FIOS, or cable modems
INFO 330
Chapter 1
7
www.ischool.drexel.edu
The Internet
• The packets are defined and handled
according to protocols, most notably the
Transmission Control Protocol (TCP) and
Internet Protocol (IP)
• A protocol is a language for
communication
INFO 330
Chapter 1
8
www.ischool.drexel.edu
Protocols
• In order for it to work, both parties (e.g.
hosts, switches, etc.) need to speak the
same language oder Sie werden einander
nicht verstehen
• Some protocols use a handshake concept
or they won’t understand each other
– Like saying Hi as a greeting, special
messages are defined that request a
connection, and reply to accept the
connection
INFO 330
Chapter 1
9
www.ischool.drexel.edu
Protocols
• More formally, then, protocols define
– The format of messages (like the spelling
of words)
– The order of messages (the syntax of
sentences, or else your messages like Yoda
will sound)
• Much of understanding networking is
understanding how these protocols work
INFO 330
Chapter 1
10
www.ischool.drexel.edu
Source of Protocols
• Internet protocols are defined by the
Internet Engineering Task Force (IETF)
– The IETF was created by the Internet
Architecture Board (IAB) and also reports to
the Internet Society (ISOC)
• The Request For Comments (RFCs)
define the actual protocols
– The first RFC was dated April 1969
– As of December 2014, there are over
7400 RFCs (see RFC Index)
INFO 330
Chapter 1
11
www.ischool.drexel.edu
Internet vs Intranet
• The Internet (a proper noun, hence is
capitalized) is the public network of zillions
of computers, toasters, etc.
• An intranet (not a proper noun) is the
generic term for a local private network
that uses the same protocols as the
Internet
INFO 330
Chapter 1
12
www.ischool.drexel.edu
Type of Internet Service
• The Internet runs distributed applications
– The World Wide Web, instant messaging,
distributed games, etc. are all distributed
applications
– These applications are developed using an
Application Programming Interface (API) to
connect to the Internet
INFO 330
Chapter 1
13
www.ischool.drexel.edu
Type of Internet Service
• There are two choices for the type of
service provided by an Internet connection
– A connection-oriented, reliable service
– A connection-less, unreliable service
• Neither guarantees how fast a message
will get from host A to host B
INFO 330
Chapter 1
14
www.ischool.drexel.edu
Connection-oriented, Reliable
Service
• This establishes a loose connection
between client and server, but not to the
switches between them
• Key traits needed from this are
– Reliable data transfer – every little bit counts
– Flow control to keep from overwhelming hosts
– Congestion control to avoid Internet gridlock
• TCP provides this service (RFC 793)
INFO 330
Chapter 1
15
www.ischool.drexel.edu
Connection-less, Unreliable
Service
• This service has no handshaking – it just
sends packets of data
– Don’t know if packets ever got there
– No flow or congestion control
• Handled by the User Datagram Protocol
(UDP), RFC 768
• Use when speed is critical, such as video
conferencing or Internet telephone
INFO 330
Chapter 1
16
www.ischool.drexel.edu
The Edge of the Network
• Now we’ll examine the contents of the
Internet from the outside in – from the
“edge” to the “core”
• Hosts (end systems) can be divided into
clients and servers
– Clients are computers that request services
from Servers
– One computer (host) can be multiple clients
and servers at once (esp. in peer-to-peer
applications)
INFO 330
Chapter 1
17
www.ischool.drexel.edu
Access Networks
• To get from a host to a distant part of the
Internet, you need to pass through the
access network
• Access networks get residential, business,
and wireless users connected
• Types of connections include
– 56 kbps dial-up modem, an analog connection
over a voice phone line
• Typically get 40-42 kbps due to line noise
INFO 330
Chapter 1
18
www.ischool.drexel.edu
Access Networks
• Digital subscriber line (DSL) gives a
dedicated connection, with different
upstream and downstream rates
– DSL uses FDM
– Downstream/upstream rates are typically
values like 768k/128k, 3.0M/768k, etc.
• Business connections may use dedicated
T1 lines (1.536 Mbps), ISDN connections,
and other options
INFO 330
Chapter 1
19
www.ischool.drexel.edu
Access Networks
• Cable modems use hybrid fiber-coaxial cable
(HFC) to connect to special cable modems
– HFC is a variant on the same cable used for
cable TV service
– HFC is a shared medium – if all your neighbors
are online, your connection speed will suffer!
• Dial-up connections are only present when
needed; DSL and cable modems are always
on (we hope)
INFO 330
Chapter 1
20
www.ischool.drexel.edu
Access Networks
• Fiber to the home (FTTH) is fiber optic
Internet connection for residential use
• There are two kinds of FTTH
– Active optical networks (AONs) are switched
Ethernet
– Passive optical networks (PONs) are used by
Verizon’s FIOS service
• Typically about 100 homes share a connection
from the provider’s central office (CO)
INFO 320 week 1
21
www.ischool.drexel.edu
Wired access
• Local area networks (LANs) generally use
Ethernet for wired connections
• Ethernet speeds of 10-1000 Mbps are
common, up to 10 Gbps for servers and
routers
INFO 320 week 1
22
www.ischool.drexel.edu
Wireless Access
• Wireless devices connect through wireless
access points (base station) on a LAN
– Then the LAN uses some other access
connection to get to the Internet
• Wireless devices use the IEEE 802.11 family
of technologies
– 802.11a supports up to 54 Mbps @ 5 GHz
– 802.11b supports 5.5 and 11 Mbps @ 2.4 GHz
– 802.11g supports up to 54 Mbps @ 2.4 GHz
INFO 330
Chapter 1
23
www.ischool.drexel.edu
Why Does Frequency Matter?
• Wireless signals can be interfered with by other
devices; when that occurs, they detune their speed
– 802.11a has seven (48, 36, 24, 18, 12, 9, and 6 Mbps)
– 802.11b has three lower data rates (5.5, 2, and 1 Mbps)
– 802.11g has a range of lower speeds
• The 802.11b and 802.11g standards use the 2.4
GHz (gigahertz) frequency range
– This frequency range is used by other networking
technologies, microwave ovens, 2.4GHz cordless phones
(a huge market), and Bluetooth devices
• The 5 GHz frequency range for 802.11a is relatively
clear, so it’s less likely to have interference (so far)
INFO 330
Chapter 1
24
www.ischool.drexel.edu
Wireless Network Example
Computer 1
Phone
line
DSL Modem
Gateway /
DHCP server
Hub
Computer 2
Computer 3
Or could have
Coax
Cable
Wireless
Access
Point
Cable
Modem
Laptop 1
Wireless
Repeater
INFO 330
Chapter 1
Laptop 2
25
www.ischool.drexel.edu
WiMAX
• The next generations of wireless
communication are a battle between
advanced cell technologies (3G and 4G
protocols) and WiMAX
• WiMAX is IEEE 802.16, and promises 510 Mbps speed over ranges of tens of km
INFO 320 week 1
26
www.ischool.drexel.edu
Physical Media
• Physical media used for connecting
networks can be guided or unguided
– Guided media use something solid – wires,
coaxial cable, fiber-optic cable, etc.
– Unguided media use electromagnetic waves
of some kind – wireless LAN signals, satellite
channels, etc.
INFO 330
Chapter 1
27
www.ischool.drexel.edu
Physical Media
• Specific kinds of physical media include
– Twisted pair copper wire
– Coaxial cable
– Fiber optics
– Terrestrial radio channels
– Satellite radio channels
INFO 330
Chapter 1
28
www.ischool.drexel.edu
Twisted pair copper wire
• Most common physical medium, has
multiple coated wires wrapped around
each other
– Includes phone lines, which have four thin
wires with RJ-11 plugs on the end
– Ethernet cables have eight wires, and RJ-45
plugs on the end, so they’re wider than phone
plugs
• Can handle Gbps speeds over distances
of about a hundred yards
INFO 330
Chapter 1
29
www.ischool.drexel.edu
Copper wire
and cylinder
Coaxial cable
Insulation
• Coaxial (coax) cable has a copper wire
core, and a copper cylinder around it –
they share the same axis of rotation,
hence the name
• Handles multiple Mbps speeds for miles
• There are only two conductors, which is
why it’s a shared medium – everyone
shares the same resources
INFO 330
Chapter 1
30
www.ischool.drexel.edu
Fiber optics
• Fiber optics use hollow fibers to guide
light pulses
• Handles hundreds of Gbps speeds up
to 100 km
• Most international phone lines, and the
Internet backbone, are fiber optic cables
• Used on high speed LANs – 1 to 10 Gbps
INFO 330
Chapter 1
31
www.ischool.drexel.edu
Terrestrial radio channels
• These include the wireless network
channels discussed previously, plus radio
signals used to beam networks between
buildings
• Can reach long distances with the latter,
but signals can be intercepted, bounce,
fade, and have interference from other
signals
INFO 330
Chapter 1
32
www.ischool.drexel.edu
Satellite radio channels
• Consist of geostationary satellites and
low-altitude satellites
– Geostationary satellites hover 24,000 miles
above the Earth’s surface, and are used to
relay TV channels and parts of the Internet
backbone
– Low altitude satellites (LEO, low-Earth
orbiting) orbit much faster, so you need
several to be able to find one at any given
time; are not used for networks
INFO 330
Chapter 1
33
www.ischool.drexel.edu
Psst – what Internet Backbone?
• The Internet is a network of many networks
– It was designed that way to be redundant in the
event of war – if one part of it was no longer
usable (nice euphemism!), the rest of the network
would still work
• At its heart are many Tier-1 ISPs
– Sprint, MCI, WorldCom, AT&T, etc. are all Tier-1
– They run extremely fast “backbone” connections
(622 Mbps to 10 Gbps)
INFO 330
Chapter 1
34
www.ischool.drexel.edu
Internet Backbone
• The Tier-2 ISPs are regional or national in
scope, and connect to Tier-1 and Tier-2
ISPs
• Points where ISPs connect to each other
are Points Of Presence (POPs)
– Don’t confuse with Post Office Protocol (POP)
• They may also connect at Network Access
Points (NAPs) to local telecom companies
or Tier 1 ISPs
INFO 330
Chapter 1
35
www.ischool.drexel.edu
Internet Backbone
• There are thousands of lower level ISPs,
Tier-3, probably including your local ISP
• For a packet to get from one host to
another, it may pass through a variety of
Tier-1, Tier-2, and Tier-3 ISPs, NAPs,
POPs, etc.
INFO 330
Chapter 1
36
www.ischool.drexel.edu
Circuit vs Packet Switching
• In order to get a packet from host A to host
B, two major approaches could be used
– Both approaches send packets over
communication lines
– Circuit switching is what a traditional
telephone system does
• Reserve a path from A to B which is the circuit
messages will follow, until the connection is closed
– Packet switching is used by the Internet
• Dump packets into the network with no reserved
path, and make a best effort to get packet to
destination
INFO 330
Chapter 1
37
www.ischool.drexel.edu
Circuit Switching
• To link host A and host B, each link
between switches along the way must
be reserved for the duration of that
connection or circuit
• There are two ways to share links with
many circuits:
– Frequency-division multiplexing (FDM)
– Time-division multiplexing (TDM)
INFO 330
Chapter 1
38
www.ischool.drexel.edu
FDM and TDM
• FDM acts like FM radio – it divides the link
by frequency ranges, and assigns a
frequency range to each circuit
– Typical frequency range, or bandwidth, is 4
kHz
– This way one link can handle many circuits
• TDM breaks the link into some number (n)
of slots in a frame
– Each slot is dedicated to one circuit, so that
circuit has full attention of the link 100/n
percent of the time
INFO 330
Chapter 1
39
www.ischool.drexel.edu
Bits and Bytes
• To review basic computer units
– A bit is a binary digit – a 0 or 1
– Typically eight bits are a byte, the shortest
word
• Old ASCII text files may use seven bits per byte,
so there are 27 = 128 ASCII characters
– Transmission rate of data is given in bits per
second (bps), or thousands or millions or
billions of bits per second (kbps, Mbps, Gbps)
– Data transfer = rate * time
• Which has units of: bits = bits/sec * sec
INFO 330
Chapter 1
40
www.ischool.drexel.edu
Key conversion point
• In dealing with prefixes k, M, G, etc., in
computer science they represent 2^(n*10)
– k = 2^10, M = 2^20, G = 2^30, etc.
• For our purposes, treat prefixes as their
base 10 equivalents
– k = 1000, M = 1,000,000, G = 1 billion
INFO 320 week 1
41
www.ischool.drexel.edu
TDM Example
• Suppose you have a 1.536 Mbps TDM
connection, and want to send a 1 Mb
(megabit) file; the connection has 12 links
• How long does it take?
– Your transmission speed is 1/12 of the
1.536 Mbps, or 0.128 Mbps
– Time = data / rate = 1 Mb / 0.128 Mbps =
7.8125 seconds
– This doesn’t include time to make the
connection
INFO 330
Chapter 1
42
www.ischool.drexel.edu
Packet Switching
• Messages are divided into packets before
going into the network
• Most packet switches must receive an
entire packet before forwarding it to the
next switch
– This store-and-forward transmission
introduces delays while the switch waits for
the entire packet to get there
• If a packet size is L, and the transmission rate is R,
the delay to receive one full packet is L/R
INFO 330
Chapter 1
43
www.ischool.drexel.edu
Store and Forward Delay
• Assume 1) no queuing delay, 2) no time to
make a connection, and 3) no delay to
propagate packets
• Send a packet of L bits across a packetswitched network with Q links, all of which
have a transmission rate of R bps
– For each link, the store and forward delay of
L/R seconds; this occurs Q times, for a total
delay of Q*L/R seconds
INFO 330
Chapter 1
44
www.ischool.drexel.edu
Packet Switching
• Each switch typically connects to many
links
• For each link, there is an output buffer (or
output queue) to hold packets waiting to
go on that link
– This introduces queuing delays, while a
packet waits its turn
– If the buffer is full, the packet can be lost –
packet loss isn’t good!
INFO 330
Chapter 1
45
www.ischool.drexel.edu
Statistical Multiplexing
• Compare circuit to packet switching
• Suppose users are active 10% of the time,
sending 100 kbps of data, and not using
the connection the other 90% of the time
• If there’s a 1 Mbps connection available:
– TDM circuit switching would need 10 slots to
allow each user 100 kbps
INFO 330
Chapter 1
46
www.ischool.drexel.edu
Statistical Multiplexing
– Packet switching could handle 35 users total
because the total number of active users will be
11 or more only 0.04% of the time (look up the
binomial distribution for details)
• The remaining 99.96% of the time, the total data rate is
less than the 1 Mbps capacity of the connection
• Hence sharing resources on demand (which
is statistical multiplexing) allows the same
performance 99.96% of the time, for over
three times the number of users!
INFO 330
Chapter 1
47
www.ischool.drexel.edu
Packet-Switched Networks
• There are two major kinds of packetswitched networks – datagram networks
and virtual-circuit networks
• A datagram network forwards packets
according to the host destination
address
– Hence the Internet is a datagram network
– Routers forward packets to make a best effort
to get them to the destination address
INFO 330
Chapter 1
48
www.ischool.drexel.edu
Virtual Circuit Networks
• A virtual circuit network forwards packets
according to virtual circuit numbers
– A virtual circuit (VC) is an imaginary connection
between the source and destination hosts
• Examples are X.25, frame relay, and asynchronous
transfer mode (ATM)
– Each packet has a VC identifier (VC ID)
– Each packet switch indexes its VC translation
table, and forwards the packet to the right
outbound link
INFO 330
Chapter 1
49
www.ischool.drexel.edu
Virtual Circuit Networks
– A key difference between datagram and VC
networks is that VC networks have to
maintain state information about
connections
• Each new VC means a new entry has to be added
to the VC translation table, and then is removed
when the connection is ended
– It also needs to keep a table to map VC
numbers to output interface numbers
INFO 330
Chapter 1
50
www.ischool.drexel.edu
Datagram Networks
• Datagram networks are like the post office
• The contents of a message (like a letter or
box) are only seen by the sender and
recipient (we hope), and in between them,
the postal service only looks at the
recipient’s address, e.g. my address is:
– 306 Rush Hall
3141 Chestnut St
Philadelphia, PA 19104 USA
INFO 330
Chapter 1
51
www.ischool.drexel.edu
Datagram Networks
– If a letter is mailed to me from outside the USA,
the first thing they need to know is that the letter
needs to go to America
– Then a machine reader finds the zip code, and
knows the letter needs to go to Philadelphia,
since 19104 is entirely within Philly
– The local letter carrier recognizes 3141 Chestnut
St as the central location for all Drexel mail
– Someone within Drexel knows where 306 Rush
Hall is, and carries the letter there
INFO 330
Chapter 1
52
www.ischool.drexel.edu
Datagram Networks
– And the receptionist in 306 Rush Hall knows
that I’m full time faculty, and puts the letter in
my mailbox
• Each step along the way, the letter is
routed essentially by reading the address
backward (USA - 19104 – Philadelphia,
PA – 3141 Chestnut St – 306 Rush Hall –
Jennifer Booker)
• Datagram networks do the same thing – a
packet of data is wrapped in layers of
addresses, which are used by routers
INFO 330
Chapter 1
53
www.ischool.drexel.edu
Datagram Networks
• Notice that datagram networks do not
maintain state information about any
packet – they only read the address and
decide where to send it based on that
address
• Traceroute (in Windows, tracert; see also
RFC 1393) is an application that shows
you the details of how a packet gets from
one host to another
INFO 330
Chapter 1
54
www.ischool.drexel.edu
Traceroute Output
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
FROM www.adelphiacom.net TO www.nero.com.
traceroute to www.nero.com (62.93.192.11), 64 hops max, 44 byte packets
1 i0.chi75.adelphiacom.net (66.109.10.17) 0.554 ms 0.420 ms 0.360 ms
2 g1-01-02-00.a0.chi75.adelphiacom.net (66.109.3.17) 0.561 ms 0.873 ms 0.313 ms
3 a1-00-00-00.c0.chi75.adelphiacom.net (66.109.3.1) 0.372 ms 0.355 ms 0.317 ms
4 so-00-01-00.c1.dca91.adelphiacom.net (66.109.0.82) 16.992 ms 16.940 ms 16.925
ms
5 p3-05-00-00.p0.dca90.adelphiacom.net (66.109.1.142) 17.748 ms 17.743 ms 17.740
ms
6 so-4-0-0.mpr2.iad5.us.above.net (64.124.11.225) 17.817 ms 17.812 ms 20.384 ms
7 so-7-0-0.mpr2.iad1.us.above.net (64.125.28.13) 17.832 ms 17.917 ms 17.798 ms
8 so-6-0-0.cr2.dca2.us.above.net (64.125.27.210) 18.178 ms 18.202 ms 18.211 ms
9 so-6-0-0.cr2.lhr3.uk.above.net (64.125.27.166) 90.064 ms 90.101 ms 97.132 ms
10 64.125.27.221.available.above.net (64.125.27.221) 107.404 ms 107.474 ms 107.519
ms
11 pos-9-1.mpr2.fra1.de.above.net (64.125.23.253) 113.379 ms 113.830 ms 113.340 ms
12 ge-9-7.er2a.fra1.de.above.net (64.125.23.186) 154.871 ms 117.584 ms 117.607 ms
13 62.93.192.11.insoft.fra2.de.mfnx.net (62.93.192.11) 113.757 ms 113.659 ms 113.576
ms
INFO 330
Chapter 1
55
www.ischool.drexel.edu
Traceroute Output
• Each line of output gives you
–
–
–
–
The hop number (1, 2, …)
The name of the server it’s passing through
The IP address of that server (e.g. 66.109.1.142)
And times of three attempts to “ping” that server
(say Hi to it), given in milliseconds (ms)
• Notice the example goes through servers in
the UK and Germany (uk, de), and the ping
times go over a hundred milliseconds
INFO 330
Chapter 1
56
www.ischool.drexel.edu
Summary of Network Types
Telecommunication
Networks
Circuit-switched
networks
FDM
INFO 330
Packet-switched
networks
TDM
Virtual-circuit
Networks
(X.25, frame
relay, ATM)
Chapter 1
Datagram
Networks
(Internet)
57
www.ischool.drexel.edu
Delays and Losses
• We’ve hinted at several kinds of things that
can delay a packet or make it get lost; now
we’ll examine those concepts in more detail
• After a packet leaves the host, it typically
passes through several routers before
getting to its destination
• Each router examines the packet’s header
to determine which outbound link it needs
to follow, and puts it in a queue for that link
INFO 330
Chapter 1
58
www.ischool.drexel.edu
Delays and Losses
• Four main causes of delay at each router:
– Nodal processing delay
– Queuing delay
– Transmission delay
– Propagation delay
transmission
A
propagation
B
INFO 330
nodal
processing
Chapter 1
queueing
59
www.ischool.drexel.edu
Delays and Losses
– Nodal processing delay is the time needed for
the router to examine the packet’s header and
choose the right outbound link
• Also may include time for error checking the packet
• Typically in microseconds for good routers
– Queuing delay is the time for a packet waiting to
be transmitted across the outbound link
• Depends mostly on how much traffic got to the router
which is waiting for the same link
• Could be microseconds or milliseconds in duration
INFO 330
Chapter 1
60
www.ischool.drexel.edu
Delays and Losses
– Transmission delay is like the store-andforward delay mentioned earlier; it’s the time
to transmit the packet onto the link
• The entire packet has to be pushed onto the link
by the router, so the transmission delay is L/R, or
(packet size)/(transmission speed)
– Propagation delay is the time for the packet
to get to the next router
• Distance = speed * time, so the propagation delay
is distance/speed, where speed is 2 or 3x108
m/sec (the speed of light is 3x108 m/s)
INFO 330
Chapter 1
61
www.ischool.drexel.edu
Delays and Losses
• You might think of Transmission delay and
Propagation delay as being like leaving for
a trip – transmission delay is the time to
pack the car (time to get out of the house),
and propagation delay is the time to drive
to your destination (travel time)
• Or ignore this analogy if it doesn’t help 
INFO 330
Chapter 1
62
www.ischool.drexel.edu
Delays and Losses
• So the delay at one node, the nodal delay,
is given by
dnodal = dproc + dqueue + dtrans + dprop
– Where dproc = Nodal processing delay
dqueue = Queuing delay
dtrans = Transmission delay
dprop = Propagation delay
– The relative magnitude of these terms can
vary widely, depending on the circumstances
INFO 330
Chapter 1
63
www.ischool.drexel.edu
Traffic Intensity
• Consider if all packets were the same size
L bits, and arrive at a router at a rate of ‘a’
packets per second
– The rate of data arriving at the router is L*a
bits per second
• The output rate from the router is its
transmission rate, R bits per second
• The traffic intensity is L*a/R
– Want traffic intensity < 1 – why?
INFO 330
Chapter 1
64
www.ischool.drexel.edu
Traffic Intensity
• Average queuing delay grows exponentially
as traffic intensity approaches one
– This is the router equivalent of gridlock!
• It was assumed that the router could hold an
infinite amount of packets in its queue
– A dropped or lost packet occurs when a packet
arrives at a router with its outbound link queue full
– Fraction of lost packets is a key measure
INFO 330
Chapter 1
65
www.ischool.drexel.edu
End-to-end Delay
• So far we focused on one router
• Now consider the total delays getting from host to
host – the end-to-end delay
• If we assume
– 1) there are N-1 routers between hosts,
– 2) queuing delays are negligible, and
– 3) processing delays are the same for each router and the
source host,
– 4) transmission rates are all R bits/sec, and
– 5) propagation delays are all equal
INFO 330
Chapter 1
66
www.ischool.drexel.edu
End-to-end Delay
• Then the total delay from source host to
destination host is
dend-end = N*(dproc + dtrans + dprop)
– And dtrans is L/R, with L the packet size
• So why is it N instead of (N-1)?
INFO 330
Chapter 1
67
www.ischool.drexel.edu
Internet Throughput
• Much of the Internet core has more
capacity than currently needed (it is overprovisioned)
• As a result, the limit of getting data
through the Internet is the speed of your
access link (ISP connection) and your
destination’s access link
INFO 330
Chapter 1
68
www.ischool.drexel.edu
Layered Architecture
• As hinted at in the syllabus, the layers of
networking are a key concept
– Why use layers?
• To solve a big problem, break it into little problems
• Each layer has a small, focused amount of work it
needs to accomplish; each layer provides services
to the layer above it
• Disadvantages are: possible duplication
of work (error recovery on multiple layers),
and violating the scope of a layer’s
services
INFO 330
Chapter 1
69
www.ischool.drexel.edu
Layered Architecture
• The layers are seen at right
– The application layer is where
user-visible software exists –
HTTP, SMTP, FTP, etc.
protocols
– The transport layer is home to
the TCP and UDP protocols
– The network layer is home to the
Internet Protocol, IP, and the
protocols used by routers
INFO 330
Chapter 1
application
transport
network
link
physical
70
www.ischool.drexel.edu
Layered Architecture
• The link layer is key for local routing;
includes Ethernet and Point-to-Point
Protocol (PPP)
• The physical layer moves the bits of data
(frames, as we’ll see shortly) across the
guided or unguided media discussed
earlier
– Each medium has protocols for how data
is encoded and decoded
INFO 330
Chapter 1
71
www.ischool.drexel.edu
But Wait Professor Booker!
• Aren’t we missing the Presentation and
Session layers?!?
– Yes, the OSI reference model has them between
the application and transport layers, but they
aren’t directly relevant here
• The presentation layer includes coding and conversion
functions that are applied to application layer data –
such as MPEG, QuickTime, JPG, GIF, TIFF
• The session layer opens and closes communication
sessions; AppleTalk is a familiar protocol here
INFO 330
Chapter 1
72
www.ischool.drexel.edu
Layered Architecture
• To make it more confusing, the packet we’ve
been discussing has different names as it
descends the layers
– Terms may vary from vendor to vendor
• A packet becomes
–
–
–
–
A message in the application layer
A segment in the transport layer
A dataframe (or datagram) in the network layer
A frame in the link and physical layers
INFO 330
Chapter 1
73
www.ischool.drexel.edu
Layered Architecture
• With each layer, headers are added to the
message to describe the address
information needed by that layer
• This process is called encapsulation, as
we put the message in bigger and bigger
boxes
• Routers and switches typically look at the
link or network layer information
– Like a letter carrier, they don’t read your mail
INFO 330
Chapter 1
74
www.ischool.drexel.edu
Layered Architecture
INFO 330
Chapter 1
75
www.ischool.drexel.edu
Network Security
• While security is covered in detail in INFO
331, we’ll mention a couple of key
concepts
• Malware is a generic term for software that
does harm (malicious software)
– It could enroll your computer in a botnet,
where it helps distribute spam or help attack
other computers
– Much malware is self-replicating, so it can
spread very quickly
INFO 330
Chapter 1
76
www.ischool.drexel.edu
Network Security
• Viruses are malware that require the user
to activate it somehow, but it could be
disguised as a web link
• Worms can enter your computer without
user activation
• Trojan horses enter via a legitimate
application, such as a simple game
INFO 330
Chapter 1
77
www.ischool.drexel.edu
Network Security
• Threats can keep a host from getting
legitimate network traffic – this is a denial
of service (DoS) attack
– Types of DoS attacks include exploiting a
vulnerability in the OS or an application,
flooding the bandwidth leading to the host, or
making the host establish phony network
connections
• Herds of computers can participate in a
distributed DoS attack (DDoS)
INFO 330
Chapter 1
78
www.ischool.drexel.edu
Network Security
• Network data can be read using packet
sniffers
– We’ll use one for our labs, WireShark
• Or people can fake who they are on the
network, and impersonate you (IP
spoofing) or intercept a network
connection (man in the middle attack)
INFO 330
Chapter 1
79
www.ischool.drexel.edu
A Little History
• The concept of packet switching was
developed in the early 1960’s by MIT and
the Rand Institute, in order to make it
possible to share really expensive
computer time efficiently
• The first packet switches were called
interface message processors (IMPs)
• ARPAnet, the Internet predecessor, was
proposed in 1967
INFO 330
Chapter 1
80
www.ischool.drexel.edu
A Little History
• By 1969, four computers were on ARPAnet,
and RFCs were being published
• By 1972 there were 15 nodes on ARPAnet,
and it was first seen publicly
• The first email program was written in 1972
• A microwave network was developed in
Hawaii, and various packet switching
networks were developed by the mid 1970’s
INFO 330
Chapter 1
81
www.ischool.drexel.edu
A Little History
• As the number of similar networks grew,
connecting them to aid researchers
became an obvious direction
• Vint Cerf helped establish the core Internet
protocols by the end of the 1970’s – TCP,
IP, and UDP
• Robert Metcalfe defined Ethernet in 1976
• By 1983, ARPAnet switched to TCP/IP
INFO 330
Chapter 1
82
www.ischool.drexel.edu
A Little History
• The French installed Minitel, a public
packet-switched network, in the early
1980’s, a decade before the US caught on
to the Internet
• DNS wasn’t invented until the late 1980’s
(RFC 1034)
• The World Wide Web was invented
between 1989 and 1991 by Tim BernersLee, based on work as far back as 1945
INFO 330
Chapter 1
83
www.ischool.drexel.edu
A Little History
• At the end of 1992 there were ~200 web
servers in the world
• In 1994 Mosaic was formed, later known
as Netscape, and much of the world was
introduced to the Internet
• By the late 1990’s, peer-to-peer file
sharing, instant messaging, email, and the
Web formed the ‘killer apps’ that launched
the world we see today
INFO 330
Chapter 1
84
www.ischool.drexel.edu
A Little History
• The dot-Com bubble burst by 2001, but a
few companies survived
• Through the 1990’s, issues such as
security and handling of streaming video
became urgent, as e-commerce became
as common as a 7-11
• Now more devices are connected –
phones, PDAs – and we can’t imagine not
having the Internet at our disposal
INFO 330
Chapter 1
85
www.ischool.drexel.edu