CS403: Online Network Exploration

1
CS403 Online Network Exploration

• The Internet
• Spring, 2007
• Modified by Linda Kenney
• January 22, 2007

2
Defining the Internet

• What is the Internet?

3
Defining the Internet (cont.)

• There are two equally valid ways to answer the
  question What is the Internet?

4
Defining the Internet (cont.)

• One perspective would be an answer based on what
  the Internet is.
• This answer would describe the physical presence
  of the Internet.
• This is the answer you might get from a computer
  professional.

5
Defining the Internet (cont.)

• A more common perspective would be an answer
  focused on what the Internet does.
• Or, more accurately, what the Internet allows us
  to do.
• This is the answer youd most likely get from a
  knowledgeable person on the street.

6
Defining the Internet (cont.)

• Well explore both perspectives, but lets start
  with a look at what the Internet allows us to do.

7
Internet services

• What the Internet allows us to do is determined
  by the services that it provides.
• When we are using the Internet, we are utilizing
  one or more of these Internet services.
• World Wide Web allows self-publication in
  electronic formats.
• E-mail allows directed communication with other
  Internet users.
• Remote login allows computers to be controlled
  remotely by users on other computers.
• File transfer allows files to be moved and copied
  between various computers.
• Network news and mailing lists allow indirect
  communication with other Internet users.

8
Data communications

• All Internet services involve the transfer of
  data between computers.
• We call this process data communication.
• Although we often think of the data being
  communicated between computers, computers are
  simply devices for running software programs.
• Nearly everything a computer does is done under
  the specific direction of a software program.
• Therefore, communications typically occur between
  software programs running on different computers.
• As you are probably aware, there are several
  different groups, companies and individuals
  writing communications software and building
  communications hardware somehow the software and
  hardware from one manufacturer must interoperate
  with software and hardware from other
  manufacturers.

9
An example

• Consider sending an e-mail as an example.
• When you send an e-mail to a friend, all you need
  to know is your friends e-mail address.
• You dont need to know what kind of computer your
  friend has, what operating system its running,
  or even what software program your friend uses
  for e-mail.
• Somehow the e-mail you are sending will get
  delivered to your friend regardless of all these
  details.
• And somehow your friends e-mail software will
  allow them to read your message.
• And somehow any reply they send will make its
  way back to you in a readable format.
• Interoperability makes this disregard for details
  possible.
• And the basis of interoperability is having a
  common ground for communications.

10
Protocols

• The languages that computers speak when
  communicating are called protocols.
• They are far more specialized, and therefore
  limited, than the natural languages spoken by
  humans.
• So, functionally, protocols serve the same
  purpose as natural language they establish a
  predetermined basis for communication.
• But practically they are far more specialized,
  dealing only with a narrow range of carefully
  determined possibilities.
• Protocols determine exactly what can be said
  and how to say it.

11
Protocols and services

• All Internet services require communications and
  all computer communications require protocols.
• Therefore, each Internet service is based upon
  one or more protocols.
• World Wide Web Hypertext Transport Protocol
  (HTTP)
• E-mail (sending) Simple Mail Transfer Protocol
  (SMTP )
• Email (receiving) Post Office Protocol
  (POP3)and Internet Message Access Protocol (IMAP)
• Remote login Telnet
• File transfer File Transfer Protocol (FTP)
• Network news Network News Transfer Protocol
  (NNTP)
• As long as hardware or software is designed to
  follow a protocol exactly, it should interoperate
  with other hardware and software based on the
  same protocol.

12
WWW Popular Activities

• What do you do on the web?

13
WWW popular activities

• From Internet Effectively A Beginners Guide to
  the WWW
• by Tyrone Adams and Sharon Scollard

14
Examples of Internet services

• http//www.rootsweb.com/jfuller/gen_mail.html
• http//www.freecycle.org/
• http//www.quiltart.com/subscribe_info.html
• http//www.ftpplanet.com/cgi/_listreview/listrevie
  w.pl?list_pathShareware_FTP

15
Future Internet Web Trends

• Continued importance of E-Commerce
• Wireless Web access
• Blogs
• Podcasting
• Wikis
• RSS (Really Simple Syndication or Rich Site
  Summary)
• Constant Change!

16
The physical Internet

• Having covered the basics of what the Internet
  allows us to do, lets turn now to an examination
  of what the Internet is.
• Stated most simply, the Internet is a
  globe-spanning collection of interconnected
  networks.
• Or stated in other words, the Internet is a
  network of networks.

17
Network basics

• A computer network is two or more computers
  connected together so that they can communicate.

18
Network basics (cont.)

• Computers connected to a network are commonly
  called hosts.
• The connection that carries data among hosts is
  called the transmission medium.

19
Growth of Internet (an aside)

• Hobbes Internet Timeline
• http//www.zakon.org/robert/internet/timeline/

Year 1969 1989 1992 1995 2001 2002 2003 2005
Host Computers 4
100,000 1,000,000 8,000,000
109,000,000 147,000,000 171,600,000353,000,000

20
Network basics (cont.)

• All the hosts on the network must somehow share
  the transmission medium.
• Only one data transfer can take place at a time
  on the transmission medium.
• This means that if host A is transferring data to
  host B, and host C needs to transfer data to host
  D, host C must wait until host A is done before
  it begins its transfer.
• If host A is transferring a lot of data to host
  B, host C could end up waiting a long time.
• For this reason, its not desirable to give host
  A exclusive access to the transmission medium
  until its done with its transfer.
• Instead networks use a system to ensure that
  access to the transmission medium is shared
  equally among all connected hosts.
• The mechanism used is called packet switching.

21
Packet switching

• Packet switching avoids network delays by
  ensuring that access to the transmission medium
  is shared equally by all hosts.
• Each host that has data to transfer is given a
  turn.
• During its turn, each host gets exclusive access
  to the transmission medium.
• However, the total amount of data its allowed to
  transfer during its turn is strictly limited.
• The largest amount of data a single host is
  allowed to transfer during one turn determines
  the size of a single packet.
• Data that is bigger than the packet size for the
  network must be broken up into multiple packets.
• If a host needs to send multiple packets, it must
  wait for later turns to send the others.
• No host is allowed to send more than one packet
  per turn.

22
Packet switching (cont.)

• Every packet is labeled with the identity of the
  host that sent it and the identity of the host
  for which its destined.
• Every host on a given network must therefore have
  a unique identity.
• Host identities are called addresses.
• Every network must have rules to determine how
  these identities are assigned.
• And every packet must be constructed and labeled
  according to a common set of rules as well.
• When we need common rules for computers to
  follow, we define protocols.

23
Packet Example

• As an example, let's look at how an e-mail
  message might get broken into packets. Let's say
  that you send an e-mail to a friend. The e-mail
  is about 3,500 bits (3.5 kilobits) in size. The
  network you send it over uses fixed-length
  packets of 1,024 bits (1 kilobit). The header of
  each packet is 96 bits long and the trailer is 32
  bits long, leaving 896 bits for the payload
  (actual message). To break the 3,500 bits of
  message into packets, you will need four packets
  (divide 3,500 by 896). Three packets will contain
  896 bits of payload and the fourth will have 812
  bits. Here is what one of the four packets would
  contain
• From http//computer.howstuffworks.com/question525
  .htm

24
Packet Example (cont.)

• From http//computer.howstuffworks.com/question525
  .htm

25
Data transfer protocols

• Each host on a network must be able to
  communicate with all the others.
• This means they must all follow a common protocol
  in order to transfer data across the network.
• These data transfer protocols set rules for how
  hosts are assigned addresses, how packets are
  formed and labeled, and all other aspects of a
  networks operations.
• The protocol is generally implemented in the
  hardware that connects each host to the
  transmission medium.
• This hardware is commonly referred to as a
  Network Interface Card (NIC).
• All hosts on a single network must therefore be
  using compatible NICs that implement a single
  networking protocol.
• We will discuss some of the specific data
  transfer protocols in use today a little bit
  later.

26
Types of networks

• There are several different types of networks.
• Each has its own protocol and other
  distinguishing characteristics.
• But all networks in widespread use today are
  fundamentally packet switching networks.
• To understand the structure of the Internet, we
  need to understand the different types of
  networks that comprise it.
• Several types of networks exist in order to meet
  a variety of needs and budgets.
• At their simplest, networks can be divided into
  two broad categories.
• Local area networks (LANs) are intended to
  connect many hosts over relatively short
  distances.
• Wide area networks (WANs) are intended to connect
  relatively few hosts over very long distances.

27
Local area networks

• The vast majority of the networks we encounter
  directly in offices, dorm rooms and computer
  clusters are LANs.
• These networks support large numbers of connected
  hosts and emphasize performance.
• But to do so, they enforce strict limits on the
  maximum length they are able to span.
• Different types of LAN have different maximum
  lengths, but they are generally limited to less
  than 1 mile.
• Consequently, LANs are in widespread use.
• LANs have become very affordable and they are
  therefore used whenever practical.

28
Wide area networks

• Most of us use WANs every day, but we seldom
  encounter them directly.
• They move data behind the scenes among smaller
  numbers of hosts over virtually unlimited
  distances.
• Since they connect fewer hosts and cover much
  longer distances, WANs generally are much more
  costly to establish and maintain than LANs.
• WANs are used only when a LAN will not work due
  to distance limitations.

29
Wide area networks (cont.)

• As with LANs, all hosts on a WAN must adhere to
  the same protocol.
• And the protocols used by WANs are different from
  those used on LANs.
• In fact, there are several mutually incompatible
  WAN protocols.
• So hosts on one WAN cannot communicate directly
  with hosts on another WAN.
• Nor can hosts on a WAN communicate directly with
  hosts on a LAN.
• Each network is inherently self-contained and
  communication is limited to the hosts on that
  network.
• Well soon see, however, that there are ways to
  overcome this limitation.

30
Data transfer protocols for LANs

• There are also several different protocols to
  govern data transfer on LANs, designed to meet
  different needs and budgets.
• Ethernet is by far the most popular protocol in
  use today.
• TokenRing is an alternative protocol that has
  become less popular in recent years.
• Fiber Distributed Data Interface (FDDI) is a
  high-performance protocol for special purpose
  LANs.
• Asynchronous Transfer Mode (ATM) is a very
  flexible protocol that carries more than just
  computer data.
• These are just a few of the many LAN protocols in
  existence.
• And for the most part they are mutually
  incompatible.
• Which means that TokenRing NICs will only work on
  a network with other TokenRing NICs.
• They cannot be intermixed on the same network as
  Ethernet NICs.

31
Topology

• Each LAN protocol is designed to operate on a
  network of a specific shape
• This shape is called the networks topology
• Different shapes require different considerations
• This is one of many reasons why different LAN
  protocols are incompatible
• The three most common topologies are bus, ring
  and star

32
Transmission media

• Many LAN protocols are designed to operate over
  specific types of transmission media.
• As with topologies, different media require
  different considerations.
• And this, in turn, results in incompatibilities
  beyond the obvious physical mismatches.
• In general, transmission media are divided into
  physical and wireless categories
• Common physical transmission media used in LANs
  include twisted pair wire, coaxial cable and
  optical fibers.
• WANs commonly use optical fiber or leased phone
  lines.
• Common wireless transmission media used in LANs
  include radio and infrared waves.
• WANs might use radio with satellites or
  microwaves with earth-based receptors.

33
Speed, bandwidth and throughput

• The relative performance of different networks is
  a very important consideration.
• Performance can be influenced by a number of
  factors, including the protocol, the topology,
  and the transmission media being used.
• In addition, there are different types of
  performance to consider
• Speed is analogous to the speed limit on a
  freeway.
• Assuming all cars always travel at the speed
  limit, that limit determines the time it takes
  any given car to get from point A to point B.
• Bandwidth is analogous to the number of lanes in
  a freeway.
• The number of lanes in a freeway determines the
  maximum capacity of that freeway. The number of
  lanes determines how many cars can travel between
  point A and point B in a given time period.
• Most commonly, network performance is discussed
  in terms of throughput
• Throughput is closely related to speed and
  bandwidth, but is a more practical measure of how
  much data can be moved from point A to point B in
  a given amount of time.
• All three are measured using the same units
  kbps, Mbps, Gbps

34
Network incompatibility

• There are many sources of incompatibilities among
  different types of networks.
• Different topologies and transmission media are
  just two examples.
• As long as your needs remain modest enough for a
  single network, these incompatibilities do not
  present a problem.
• However, if you have a growing company, at some
  point you are likely to exceed the maximum number
  of hosts a single network can support or the
  maximum distance one can span.
• At this point, the incompatibilities will become
  an issue.
• How do you expand your business with more hosts
  or greater distances?
• Generally, by adding additional networks
• When you have an organization with multiple
  networks, you need to devise some way to connect
  those networks together otherwise, some employees
  cannot communicate with some of their coworkers

35
Interconnected networks

• When two or more networks are connected together
  they form what is called an interconnected
  network.
• This phrase is commonly shortened to internet.
• Note that the lowercase i is used because its
  a common noun.
• To connect two networks together, all we need is
  a host with a separate NIC for each network.
• And some software to tell that host how and when
  to transfer data from one network to the other.
• If this host sits between two networks using the
  same protocol, it is considered a bridge.
• If it sits between two networks using different
  protocols, it is considered a gateway.
• For our purposes, bridges and gateways are
  important only insofar as they function as
  routers.
• A router is a host connected to two or more
  networks that interconnects those networks and
  routes data between them.
• Routers are the glue that holds the various
  networks in an internet together.

36
Advantages of internets

• Using routers to interconnect various networks
  greatly expands the range of possibilities.
• We can use a LAN when we have to connect several
  hosts in a concentrated area.
• If we have more hosts than a single LAN can
  support, we can interconnect multiple LANs.
• If we have different groups of hosts with
  different communications needs, we can
  interconnect LANs of various types.
• If we have hosts that need to be connected over
  long distances we can use a WAN.
• If we have clusters of hosts scattered over a
  wide area we can use one or more LANs at each
  clustered site and one or more WANs to connect
  the sites.
• With internets at our disposal, our
  communications are no longer limited by distance
  or the number of hosts .

37
The Internet defined

• So whats the difference between an internet
  and the Internet?
• Really, the difference is nothing more than one
  of scale.
• The Internet (we use an uppercase I to indicate
  that its a proper noun) is nothing more than a
  really, really big internet.
• Like any other internet, the Internet relies on
  routers to connect thousands of individual LANs
  and WANs located all over the planet.
• By connecting those LANs and WANs together, the
  Internet effectively connects the millions of
  hosts that are in turn connected to those
  individual networks.

38
Redundant connections

• As internets grow in scale and complexity, its
  not unusual for redundant connections to be
  incorporated into them.
• Redundant connections allow multiple paths
  between various networks within the internet
• This increases flexibility and reliability
• The net effect creates a series of paths similar
  to the roads on a map
• Looking at most maps there are likely to be a
  variety of routes one might choose to get from
  point A to point B
• Note that a given path may involve a variety of
  network types, each with its own protocol
• Theres no way to predict which path your data
  will take
• And even if there were, it would be impractical
  to require each host to understand the protocols
  of every conceivable type of network
• In effect, data can get to any place on an
  internet, but theres no way to predict what it
  will encounter along the way

39
Combining packet switching networks

• Recall that on a packet switching network, each
  host is limited to sending only one packet per
  turn.
• And the data transfer protocol for that specific
  network determines the maximum size allowed for
  that packet.
• If theres no way to predict what types of
  networks our data will need to cross as it
  traverses an internet, we should send that data
  in a form thats likely to work with the packet
  sizes of all the networks it may need to cross.
• This works well, since each small packet travels
  independently across the internet.
• Small, independent packets are more flexible in
  how they travel and interact.
• Much like cars on highways are more flexible than
  trains on tracks.
• Cars will generally make more efficient use of a
  roads overall capacity.
• Trains spend a fair percentage of their time
  waiting for other trains to clear the track.
• If a packet doesnt reach its intended
  destination, its easy to replace just that
  packet.
• If we were trying to send all the data as a
  single large chunk, wed need to replace the
  whole thing every time something happened.
• In packet switching internets, the routers that
  interconnect the networks act like switches,
  directing the packets onto the next leg of their
  journey.

40
The need for a virtual network

• Another consequence of interconnecting networks
  of various types is that its impractical for
  each host to know the protocols used by every
  network involved.
• Ideally, it should not matter what type of
  network each host is connected to on an internet.
• What we want is for the hosts to behave as if
  theyre on a single network even though they are
  not.
• In computer science, when we want to behave as if
  something exists even though it doesnt, we use
  the term virtual to describe it.
• Therefore, in this case, what we want is for our
  internet to function as a virtual network even
  though in reality it is a collection of several
  different networks.
• With a virtual network in place, all hosts
  connected to it should behave as if they are
  connected to the same network.

41
The virtual network illustrated

• What we know is there

42
The virtual network illustrated

• What we wish was there

43
Creating a virtual network

• The essence of a virtual network is that all
  hosts connected to it can communicate equally,
  just as they would on an actual network.
• We now know that when we want hosts to
  communicate, we need them all to play by the same
  rules.
• And when we want a bunch of hosts to play by the
  same rules, we define those rules formally as a
  protocol.

44
Creating a virtual network (cont.)

• Therefore, what we need to establish a virtual
  network on an internet is a protocol that can be
  followed by all the hosts on that internet
  without violating the protocol of their specific
  network.
• In other words, our virtual network protocol
  should be applicable to a host on an Ethernet
  network in such a way that that host can still
  follow the Ethernet protocol (Since it will still
  need to use the Ethernet protocol to connect to
  its network).
• Likewise, for FDDI, TokenRing, ATM and others.

45
Internet Protocol

• The protocol that is used to turn the whole
  Internet into a single virtual network is called
  the Internet Protocol (IP).
• This protocol performs three primary tasks to
  define the operation of the virtual network.
• IP defines the rules for how datagrams are
  formed.
• IP defines the rules for how routers function to
  move the datagrams toward their destinations.
• IP defines the rules for how hosts on the network
  are uniquely identified.
• In general, IP is not implemented in hardware
  like NICs instead it is implemented in software.
• All hosts connected to the Internet must have IP
  software.
• Thats what allows the Internet to function as a
  single, enormous virtual network.

46
IP datagrams

• The virtual network created by IP is a packet
  switching network.
• That simply means that IP requires all hosts to
  break up data into small packets before sending
  it.
• When these IP packets travel over actual networks
  inside the virtual network, however, they need to
  travel inside the packets dictated by that
  network.
• For this reason, IP calls its own packets
  datagrams to create a distinction.
• IP carefully defines exactly how each host should
  build its datagrams.
• This allows all other hosts on the Internet to
  recognize any and all datagrams they receive.
• In addition to the data it carries, each packet
  is also labeled with the identity of its source
  host and the identity of its destination host .

47
IP routing

• In order to travel over the actual networks
  within the virtual network, IP datagrams must be
  placed within packets appropriate to that actual
  network.
• When they need to move between two networks, IP
  software on the router between those networks
  must extract the IP datagram from the incoming
  packet and insert it into an appropriate outgoing
  packet.
• Neither of the routers NICs ever has to deal
  with an IP datagram.
• Instead, each NIC only has to deal with the
  packets it was designed to handle.
• The IP datagram is wrapped up inside the network
  packet as data.
• And the NIC doesnt ever need to know anything
  about the data inside a packet.
• IP defines specific rules for the routers to
  follow when transferring datagrams between
  networks.
• Since all the routers follow the same rules, they
  are able to work cooperatively without
  communicating among themselves.

48
IP addresses

• As with any network, hosts on the Internet must
  each have a unique identity.
• Otherwise, thered be no reliable way to send
  packets to a specific host .
• Because computers are inherently digital, IP
  defines a system of unique numeric identities for
  hosts .
• These numeric identities are called IP addresses.
• Every host on the Internet must have its own
  unique IP address.
• In order to communicate with the IP software on a
  destination host , the IP software on a source
  host must know the numeric IP address of that
  destination.

49
IP addresses (cont.)

• Because there are so many computers on the
  Internet, IP addresses must be very large
  numbers.
• Since its hard to work with very large numbers
  accurately, IP addresses are commonly written
  using dotted quad notation.
• Dotted quad notation consists of four numbers (in
  the range 0 to 255) separated by dots (or
  periods)
• For example, 137.177.137.7 is the IP address for
  the cisunix computer named turing

50
IP versions

• Currently there are two types of Internet
  Protocol (IP) addresses in active use IP version
  4 (IPv4) and IP version 6 (IPv6). IPv4 was
  initially deployed on 1 January 1983 and is still
  the most commonly used version. IPv4 addresses
  are 32-bit numbers often expressed as 4 octets in
  "dotted decimal" notation (for example,
  192.0.32.67). Deployment of the IPv6 protocol
  began in 1999. IPv6 addresses are 128-bit numbers
  and are conventionally expressed using
  hexadecimal strings (for example,
  10800008800200C417A).
• http//www.iana.org/ipaddress/ip-addresses.htm

51
Assignment of IP addresses

• Every single computer added to the Internet must
  be given an IP address that is not already in
  use.
• With millions of computers already on the
  Internet and thousands more being added every
  day, thats a considerable challenge.
• Its not practical for each individual computer
  user to come up with their own unique IP address.
• Imagine how many youd need to try before you
  found an unused one!

52
Assignment of IP addresses (cont.)

• Since virtually all connections to the Internet
  are managed by an organization of some sort
  (school, company, ISP, etc.), its much more
  practical to place each organization in charge of
  its own addresses.
• When an organization first establishes a
  connection to the Internet, it is assigned an
  address space consisting of the first one, two or
  three values in a dotted quad address.
• Once assigned such an address space, the
  organization is free to assign the remaining
  three, two or single values as it sees fit to
  each computer it connects to the Internet.
• No other organization is entitled to create
  addresses within that assigned address space.
• And the organization that owns a given address
  space is solely responsible for ensuring that no
  address within that space is ever in use by two
  computers at the same time.
• At UNH, our address space consists of all
  addresses that begin with 132.177.

53

• Both IPv4 and IPv6 addresses are assigned in a
  delegated manner. Users are assigned IP addresses
  by Internet service providers (ISPs). ISPs obtain
  allocations of IP addresses from a local Internet
  registry (LIR) or national Internet registry
  (NIR), or from their appropriate Regional
  Internet Registry (RIR)
• http//www.iana.org/ipaddress/ip-addresses.htm

54
How datagrams travel

• So how does all this fit together to get
  datagrams from their source to their destination?
• The IP software on the source host creates a
  datagram labeled with the IP address of the
  destination host and gives it to the NIC as data.
• The NIC on the source host creates a network
  packet that contains the datagram, labels it with
  the network address of the router on its network
  and sends that packet out across the network.
• The NIC connecting the router to the source
  network receives the incoming network packet,
  extracts the datagram and passes it to the IP
  software on the router.
• The IP software on the router examines the
  destination IP address in the datagram and passes
  it to whichever of the routers NICs will move it
  closer to that destination.
• The outgoing NIC on the router wraps the datagram
  in a network packet suitable for the outgoing
  network, labels it either with the network
  address of the destination host or another router
  on that network (if the destination host is not
  on that network) and sends it out on that
  network.
• If the network packet is sent to another router,
  that router repeats the above three steps, moving
  the datagram on the next hop of its journey
  until it eventually reaches the router that can
  deliver it to the destination host.
• When the network packet eventually reaches the
  NIC on the destination host, that NIC extracts
  the datagram from the network packet and passes
  it to the IP software running on the destination
  host.

55
Datagrams are mindless

• In other words, datagrams are mindless and
  passive.
• They do not actively seek out their destination.
• In fact, they have no abilities whatsoever.
• Like letters in the mail, they rely on others to
  move them along on their journey to their
  destination.
• Letters have an address on the envelope and
  postal employees use that address to move the
  letter progressively closer to its intended
  recipient.
• Datagrams are labeled with the IP address of
  their destination host and the routers of the
  Internet move the datagram progressively closer
  to its intended destination.
• And like the postal service, IP is what we call a
  best effort system.

56
IPs shortcomings

• Like the U.S. Postal Service, IP is remarkably
  good at what it does, but things can go wrong.
• As with letters sent by first class mail,
  datagrams are not guaranteed to reach their
  destination.
• The overwhelming majority of them do, but for any
  single datagram theres always a small chance it
  will disappear in transit.

57
IPs shortcomings (cont.)

• There are a variety of reasons why a datagram may
  disappear en route, but they generally all have
  to do with the fact that IP was designed (like
  the postal service) to favor efficiency over
  reliability.
• It endeavors to deliver as many datagrams as
  possible as quickly as possible, but if some get
  lost in the process so be it.
• The most notable obstacle to efficiency on the
  Internet is congestion.
• If you view datagrams as analogous to cars,
  network congestion would be equivalent to a
  traffic jam on a freeway.
• The problem with this analogy is that datagrams
  traveling on a network cant stop and sit their
  until the traffic jam clears.
• Instead, when congestion occurs on an Internet,
  the routers surrounding the congestion begin
  discarding datagrams and/or directing incoming
  datagrams along an alternate route.
• This can cause the occasional disappearance, or
  dropping, of datagrams.
• It can also cause datagrams to arrive out of
  order or even in duplicate.
• Fortunately, theres a protocol that addresses
  this issue.

58
Transmission Control Protocol

• The protocol that is commonly used with IP to
  provide guaranteed delivery of datagrams is
  Transmission Control Protocol (TCP).
• TCP was designed in concert with IP.
• IPs shortcomings are not problems, per se
  they are unavoidable consequences of IPs
  priorities.
• Sometimes the loss of some packets is not a
  problem.
• But when it is, TCP can be layered above IP to
  ensure that datagrams arrive at their destination
  in order without loss or duplication.
• Since TCP is layered over IP, a program wishing
  to send data across the network would pass that
  data to the TCP software and TCP would then pass
  it to IP which would then pass it on to the NIC
  for transmission.
• TCP stamps each outgoing datagram with a
  sequential number so the TCP software on the
  receiving host can reassemble the data in the
  proper order and identify datagrams that are
  missing or have been duplicated.
• Any missing datagrams can then be requested from
  and resent by the source host.
• The TCP software on the source and destination
  host communicate with each other to ensure the
  transmission goes smoothly.
• This allows the source and destination host to
  carry on a two-way communication quite easily.
• Because of this communication back and forth, TCP
  is considered a connection-oriented system.

59
Hostnames

• Having learned the importance of IP addresses,
  its natural to wonder why we dont see them more
  often.
• Logically, one might think that an IP address
  would be necessary for any communications on an
  internet, be it e-mail, web browsing, etc.
• We dont see IP addresses very often because
  theyre generally meant for use by software.
• As humans, we generally dont like to deal with
  large numbers any more than we have to.
• Instead, we prefer to give things names, and
  hosts are no exception.
• So in addition to having a unique IP address,
  nearly all hosts on the Internet are also
  assigned a unique hostname.
• IP addresses are necessary for software, but
  hostnames are more convenient for humans.

60
Assignment of hostnames

• Since hostnames, like IP addresses, establish
  identities for hosts, they too must be unique
  across the entire Internet.
• Recall that with IP addresses, an organization is
  assigned a unique address space when it connects
  to the Internet.
• At the same time, an organization also typically
  obtains a unique domain name.

61
Assignment of hostnames (cont.)

• Domain names are an organization-specific
  identifier combined with a high-level domain
  identifier indicating the type of organization or
  the country in which its registered.
• Common high-level domains include .com, .edu,
  .net, .gov, .org, .mil, .int, .aero, .name, .biz,
  .museum, .info, .coop, .pro
• Country-specific high-level domains include .ca,
  .uk, .de, .jp.
• For example, the domain name for UNH is unh.edu.
• UNH is free to name its hosts anything it wishes,
  provided the name is used for only a single
  computer at a time and ends in .unh.edu.
• http//www.unh.edu/NIS/Docs/Intranet/index.html

62

• Welcome to the InterNIC Website!
• This website has been established to provide the
  public information regarding Internet domain name
  registration services and will be updated
  frequently.
• http//www.internic.net/

63
URL Uniform Resource Locator

• URL
• Represents the address of a resource on the
  Internet.

64
The problem with hostnames

• There are millions of hosts on the Internet, and
  humans prefer to refer to them using hostnames.
• Yet the software those humans use must have an IP
  address in order to properly address its packets.
• The dotted notation used for both might initially
  suggest that a simple translation is possible.
• But this is just a coincidence there is no
  direct correspondence between the numbers in an
  IP address and the components of the associated
  hostname.
• IP addresses always consist of four numeric
  parts, while hostnames might consist of two,
  three, four, five or more parts.
• To better understand this problem, lets use
  phone numbers as an analogy.
• To call someone, you must have their phone
  number.
• Yet, most of us are far better at remembering
  names than numbers.
• Fortunately, if we know someones full name, we
  can call directory assistance to get their phone
  number.
• What we need on the Internet, then, is something
  like directory assistance that software can use.

65
Domain Name Service

• Domain Name Service (DNS) is the Internets
  equivalent to directory assistance.
• However, its generally used by software, not
  humans.
• Humans give host identities to their software as
  hostnames.
• The software then uses DNS in order to find the
  IP address that matches that hostname.
• If the software needs to announce an incoming
  communication, it can also use DNS to find the
  hostname that matches the source IP address in
  those datagrams.
• This keeps both the humans and their software
  happy.
• When it is first connected to the Internet, a
  host is configured with the IP addresses of one
  or more hosts that will provide it with Domain
  Name Service.
• For most of us, those are the only IP addresses
  we will ever need to enter.
• From then on, we need only use hostnames.
• Any time we do, the software can contact one of
  those DNS servers to find the matching IP
  address.
• We often call this process resolving the
  address or DNS resolution.

66
The client/server model

• Note that we used the term DNS server for the
  host that our software uses to resolve addresses.
• More specifically, we use this term to refer the
  software running on that host that provides the
  services our software requires.
• In general, we call any software that provides
  services to other software a server.
• For example, a DNS server is software that
  resolves addresses as a service to other
  software.
• Any software that utilizes the services of a
  server is generally referred to as a client.
• For example, any software that uses a DNS server
  to resolve an address is acting as a client when
  it does so.
• The division of labor between programs acting as
  servers and other programs acting as clients is
  collectively known as the client/server model.
• Nearly all Internet services you use are services
  provided by servers somewhere on the Internet.
• And nearly all the programs you use on your host
  to access those services qualify as clients.

67
Summary

• The rest of this semester will involve learning
  more about various Internet services.
• Knowing how the Internet works behind the scenes
  will help us better understand those services.
• Nearly all of the services available on the
  Internet are based on the client/server model.
• And nearly all of the communications that take
  place between those clients and servers occurs
  using TCP and IP.
• In fact, the collection of protocols that any
  host connected to the Internet must implement in
  software is collectively known as the TCP/IP
  Protocol Suite.
• UDP, DNS, HTTP, SMTP and nearly all of the other
  protocols well cover this semester join TCP and
  IP in this collection.
• The first Internet service we will examine in
  depth is the World Wide Web.

68
Sample Review Questions

1. Weve discussed two equally valid answers to the
   question What is the Internet? Can you explain
   them?
2. What are Internet services? What makes them
   possible? Can you give some examples?
3. What is a protocol and why do they exist? Compare
   and contrast them with natural languages and give
   some examples of each.
4. What is packet switching and why is it useful?
5. What is the underlying physical structure of the
   Internet? Why is the Internet structured that
   way?
6. Can you define the terms data transfer
   protocol, topology, transmission media,
   speed, bandwidth and throughput? Can you
   give some examples of each?
7. What is the difference between an internet and
   The Internet?
8. What are the respective roles of the Internet
   Protocol (IP) and Transmission Control Protocol
   (TCP) and how do they fulfill those roles?
9. Can you explain the concept of a virtual network
   and its importance in the context of the
   Internet?
10. Explain the different roles played by IP
    addresses and hostnames. How are they assigned?
11. What is the role of Domain Name Service (DNS)?
12. Explain the meaning of the client/server model
    and discuss its importance to the Internet. Give
    two examples of its use.

69
Key terms

• Address
• Address space
• ATM
• Bandwidth
• Best effort system
• Bridge
• Broadcast communications
• Bus
• Client
• Client/server model
• Coaxial cable
• Computer network
• Connection-oriented system
• Data communication
• Datagram
• Data transfer protocol
• Direct communications
• DNS
• DNS resolution

Physical transmission media Protocol Radio
waves Redundant connection Remote
login Ring Router Server SMTP Speed Star TCP Telne
t Throughput TokenRing Topology Transmission
medium Twisted-pair wire UDP Virtual
network WAN Wireless transmission media World
Wide Web
File transfer FTP Gateway High-level
domain Host Hostname HTTP Infrared
waves Interconnected network Internet Internet
service Interoperability IP IP address LAN Layerin
g Mailing list Natural language Network
news NIC NNTP Optical fiber Packet Packet
switching
70

• Thanks to Mike Gildersleeve for sharing the
  information from his Summer, 2006 CS403
  PowerPoint.
• Information also used from
• Web Developer Design Foundations with XHTML by
  Terry Felke-Morris