Week 2 Lecture

1
DB3009N Week4 Part1 Background To The
Internet and WWW technologies

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
2
The problem

• In the 1950s and 1960s computers where expensive.
  Perhaps a company or university department could
  afford one. The idea of a computer per employee
  was a dream
• By the 1970s there where enough computers in a
  single company or university to make sharing data
  between them useful. This lead to Local Area
  Network technologies such as Ethernet, Token
  Ring, Token Bus, Dec net and other technologies.
• These technologies where limited to only
  connecting tens or hundreds of machines together
  with wires not more than some 100s of meters long

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
3
The problem

• People wanted to share data between sites, cities
  and countries but
• All the local Area Network technologies where
• Short range only
• incompatible with each other.
• Physically connectors, wires
• Electrically what voltage for a 1 and a 0, speed
  etc
• Protocols how and when to talk to the network
• Addressing (names of machines)
• The challenge was to build a network of networks
  The Internet

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
4
The solution - Layers

• To try and connect up such a mess of independent
  incompatible short distance products required a
  structured approach which provides hardware
  independent
• delivery of messages
• names for networks and machines.
• The solution was to split the problem up into
  layers, each layer relying on the standardised
  services of the layer bellow and each layer
  providing standardised services to the layer
  above

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
5
The Simplified Internet
Application (wants to send data between machines)
Still evolving e.g. HTTP SMTP SNMP TP TELNET
Standard format for data for email, file transfer
etc
get bytes from users program from one machine to
another (independent of application)
The internet TCP, UDP and IP
get groups bytes (PACKET) from between networks
(independent of hardware)

• get groups of bytes (FRAME) from A to B on a
  single Network
• (hardware dependant)

Existing Local Area Network technologies
get bits from onto a single network (hardware
dependant)
CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
6
The Internet

• The Internet is a network of local area networks.
• Connects millions of computers around the world
  using various technologies including phone lines,
  fibre optic lines, coaxial cable, satellites, and
  wireless connections. (You could use carrier
  pigeon)
• Normally individual users at home do not connect
  to the Internet directly instead you connect
  through a Internet service provider (ISP) by
  modem or ADSL etc.
• Larger companies and colleges may be directly
  connected to the internet
• The internet provides many useful features such
  as the ability to transfer files between
  machines, remote login into machines and transfer
  email as well as supporting the entire world wide
  web. Even Microsoft's much advertised .NET is
  just an extension to the internet. It does not
  replace it. Only hide it!

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
7
The Internet
Servers
Router
Clients
Network
Internet
Network
Network
Network
CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
8
Implementing the Internet

• The Internet is basically two layers over the
  existing local area technology.
• Internet Protocol (IP) provides addresses for
  networks and machines, routes packets from the
  source to the destination machine and can deliver
  up to 64Kbytes in a packet. There is no error
  recovery mechanism. You don't know if the packet
  arrived or not. You also can't tell if the data
  in the packet was damaged on route. There is no
  security provided. The order packets are
  delivered in is not guaranteed. Packets may
  become duplicated
• The next Layer up is basically two alternative
  systems UDP and TCP
• User Datagram Protocol provides addresses for
  applications and the ability to communicate
  between applications otherwise its like IP
  cheap and cheerful
• Transmission Control Protocol provides the
  ability for an application to send almost
  unlimited data. (it is split into IP packets and
  the order adjusted if they arrive out of order.)
  Its reliable. lost and damaged packets are resent
  duplicates are discarded. Its slower and more
  expensive than UDP. Easier to program. The world
  wide web uses TCP

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
9
Advantages of TCP/IP

• It works!
• TCP/IP provides the base on which other Internet
  services depend.
• The protocol is successful because of following
  reasons
• Efficient and general purpose it can support
  many services
• Flexible and robust it can be used with many
  underlying communication technologies (e.g. with
  Local Area Networks as well as Wide Area
  Networks)
• Its flexibility has allowed engineers to use
  TCP/P with computers, networking technologies,
  and services that did not exist when TCP/IP was
  designed.
• It withstood rapid growth of Internet since
  1990s.
• It is not been stolen or ruined by Microoft
  (yet)
• Microoft has no control over it

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
10
Structuring the relationship between clients

• The designers of the internet decided not to
  limit what applications or data could be
  transferred or the role of the clients.
• These models define the relationships between the
  clients and their roles without defining the
  clients, applications or the data

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
11
Structuring the relationship Master-Slave
Networking

• In early days, computers were large and
  expensive. Most companies could afford only a
  single computer. Only one person could use the
  machine at a time and they where very difficult
  to use and program (no GUI often almost no
  operating system and easy to break)
• Later dumb terminals where connected to the
  computer so that several users could login into
  the machine at the same time and share it. This
  required operating systems such as UNIX that
  could multi task and were also multi user (No
  windows Os is multi user)
• The dumb terminals where completely controlled by
  the main machine. This is master slave.
• OBSOLETE relationship

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
12
Structuring the relationship Distributed
Computing

• With the advent of mini computers, work stations
  and later personal computers (PCs) computers
  became so inexpensive groups or even individuals
  could have one without the need to share via dumb
  terminals.
• In the master slave model data and service
  sharing is implicit because all the data is
  actually on a single machine.
• With Distributed computing the sharing must be
  implemented using network and possibly internet
  technology because the data and users do not
  share the same machine
• The machines can act independently of each other
  and share data when they need to.
• The original goal of DS was to try to make the
  illusion that everyone shared one giant machine
  however, this has proven very very difficult to
  achieve!

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
13
Structuring the relationship Client-Server
Computing

• The goal of making all the machines look like
  one machine is so difficult a simpler model
  developed called client server
• Servers provide services to clients. This role is
  fluid. For instance a server may use the services
  of another server. I.e. a server of one service
  may be a client of another service
• In general users becomes clients of the servers.
• For example. The user wants to send an email
• User writes an email on their machine using an
  email client
• Email client sends the email to the nearest email
  server
• That email server forwards the email to another
  server and so on towards its destination

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
14
Client-Server Computing Replicating the Service

• Since a server provides a service it makes sense
  to not limit the system to serving a single
  client at a time. Especially if each request
  takes a long time or the service is popular.
• This can be achieved by running multiple copies
  of the service (replicas) of server programs in a
  computer to permit multiple users to access the
  service concurrently (simultaneously).
• Server programs usually run on large computers
  because they will receive concurrent requests and
  it makes sense to spend money on the things that
  get used a lot.

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
15
Client-Server Computing Problems

• When the server computer crashes, clients will
  not be able to send requests or receive results.
  Often the client relies upon the service and
  can't provide it itself...
• Servers handle many requests and are expensive so
  there is a tendency to add more clients (who
  often pay) to increase efficiency but each client
  will get a slower service
• The client often has little control over what
  happens after it submits a request to a server
  for a service.
• The service speed may vary at popular times it
  will be slow and less popular times it will be
  faster

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
16
Client Server System Example DNS

• Each IP packet contains the source and
  destination addresses of the two computers
  involved in the communication
• Space inside a packet is at a premium so instead
  of using text addresses like webmail.londonmet.ac.
  uk IP packets contain addresses consisting of 4
  byte numbers. E.g. 192.12.21.7
• IP therefore needs a mechanism to translate text
  names (domain names) into IP addresses. This is
  done by the Directory Name Service DNS

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
17
Client Server System Example DNS

• DNS names are organized into a tree. This is a
  common computer since solution to handling lots
  of information
• DNS is a client server based system which is
  distributed.
• DNS data is spread between servers each one
  looking after a domains like londonmet or a
  larger domain like ac.uk
• Because the data is spread out a server failure
  will only affect a domain.
• Servers may be replicated to increase performance
  and reliability
• Easier to manage since changes to a domain only
  effect that domain server and not the others
• Clients contact their local DNS server with a
  domain name to look up and if the server knows
  the address it returns the IP address to the
  client.
• If the local DNS server does not know the address
  it passes the request to the DNS server above it.
  This process is repeated until the address is
  found or the request goes back to the root of the
  tree. At the root of the tree the request is
  passed down towards the branch of the tree that
  contains the IP address of the machine

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
18
Remote domain name server in France
Local domain name server in UK
1
2
3
4
Internet
Remote computer in France
Users computer in UK
1. Application asks local domain name server for
the IP address of a computer in France 2. UK
server does not know the answer and contacts the
remote server in France automatically 3. The
remote server uses DNS to obtain the IP address
and returns to UK server 4. UK server services
the request
CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
19
World Wide Web (WWW)

• The World Wide Web is the name given to the part
  of the Internet that can accessed with the Web
  browsers (e.g. Netscape).
• It was invented by Tim Berners-Lee (English) of
  CERN (The English also invented the first
  mechanical (Charles Babbage) and first electronic
  computers (ENIGMA))
• The Web consists of millions of web sites (e.g.
  yahoo.com, microsoft.com, bbc.co.uk) comprising
  billions of web pages linked to each other and
  provides access to diverse and rich information
  resources.
• A web page is written in the markup language
  HTML. A page may include many types of media
  including text, graphics, animations, audios
  videos etc.
• Web pages are linked together using hyperlinks.
  When a hyperlink is clicked on the web browser
  loads the target page.
• Links work even if the target page is on a
  different machine from the source.
• However, since the links contain the exact
  location of he page, if the page moves the link
  needs updating. There is no mechanism to
  automatically perform this update
• Modern web pages can be static (read from a file)
  or dynamically created (requires a program to be
  run to generate the page)

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
20
Yet More Layering the WWW

• The world wide web is built on top of the
  internet on top of the client server model.
• The job of the WWW layer is to transfer a web
  page from the web server to the web browser.
• The web page is structured into HTML (Hyper Text
  Modelling Language)
• HTTP Hyper Text Transfer Protocol was established
  to structure the transfer for HTML provide error
  messages etc.
• URLs Universal Resource Locators where created to
  give each web page a name

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
21
The Simplified WWW
Web browser wants a page from a web server
This is the World Wide Web
Pages structured into HTML and referenced via URLs
HTTP used to structure communication between
client and server
get bytes (in this case HTTP packets) from users
program from one machine to another
The internet TCP, UDP and IP
route groups of bytes (IP PACKET) between
networks (independent of hardware)
get groups of bytes (FRAME) from A to B on a
single Network (hardware dependant)
Existing network technologies
get bits from A to B on a single
network (hardware dependant)
CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
22
Web Browsers

• A browser is a Web application that uses HTTP
  protocol to interact with the Web server.
• It obtains files and data requested by the user
  through the server from remote sites.
• The location of the page is written as a URL
• It renders (formats and displays) text and
  graphics in the browser window.
• The look of the page is determined by the HTML
  commands within the page
• Some popular browsers
• Netscape
• Internet Explorer

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
23
Uniform Resource Locator (URL)

• URL specifies the location of the page to be
  obtained from a web server. The page is usually
  displayed or some times simply downloaded as a
  file onto local disk.

http//www.unl.ac.uk80/simt/index.html
CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
24
Hyper Text Transfer Protocol

• HTTP is the communication protocol (rules) built
  on top of TCP IP to allow web servers and web
  browsers to communicate together
• HTPP has several interesting design decisions
  which will have effects on the way we can program
  web applications.
• HTTP does not store any information between
  requests!
• Cannot get a history of client requests
• This makes programming a shopping basket
  impossible because you can't remember what is in
  it.
• Cannot impose rules such as you must add items to
  the basket before you checkout

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
25
Hyper Text Transfer Protocol

• Why make HTTP stateless if it causes us so many
  problems?
• The designers originally intended web pages to be
  read from files and not created on the fly by
  programs so no history was needed
• Error recovery is really simple just restart the
  server there is no messy chains of requests to
  complete
• Makes programming the web server and web client
  easier.
• Many internet services (UNIX such as file
  sharing) are stateless for these very reasons
• Extensions to the web browsers and web servers
  (sessions and cookies) gets around this problem

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
26
HTTP Protocol Part 1

• Web page serving is a classic client server
  system.
• The client web browser opens a TCP connection
  with the server specified in the URL the user
  typed.
• By default this connection initially connects to
  port 80. (the client can use any port. The IP
  packet will contain the clients port number so
  the web server will know which port to send to)
• TCP automatically creates a temporary port on the
  web server for the rest of the conversation
  leaving port 80 free for another client to
  connect to...

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
27
HTTP Protocol Part 2

• After the initial connect the client sends a URL
  request for a single web page or item on a web
  page to the web server
• The web server sends the client the contents of
  the page or item
• If the web server can not find the item an error
  message is returned instead
• When the conversation is over the connection is
  closed allowing the temporary port to be reused
  if necessary
• This type of communication is called request
  response because the client sends a single
  request and gets a single reply.
• However, this is inefficient because a connection
  has to be opened and closed for every page
  fetched

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
28
Hyper Text Transfer Protocol

• Client opens connection to server on a port 80
  (TCP assigns a temporary port for the duration of
  the conversation on the server)
• Client sends request with URL identifying
  resource required
• Server sends the client the resource or an error
  message
• The client closes the connection and the
  temporary port is de-allocated

Server machine
Client Machine
Port 80
1. Connection open
Port ????
Web Server application
Temporary Port
2. request
Web Server application
3. reply
4. close
CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
29
HTTP Get request

• GET /im269/w2.html HTTP/1.1
• Extra stuff like cookies
• (much more about these later)...

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
30
HTTP Reply
HTTP/1.1 200 OK Server Apache/1.3.9
(Unix) Last-Modified Mon, 22 Sep 2003 012005
GMT Content-Length 3554 Content-Type
text/html lthtmlgt ltheadgtlttitlegt im269 Lecture
Notes lt/titlegtlt/headgt ltbodygt im269 Course
Contents. lt/bodygt lt/htmlgt
CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)
31
What you should have learnt

• Some idea of how the internet works
• Jobs of TCP, UDP, IP and DNS
• The importance of layers
• The details of how web pages are fetched
• Request reply protocol
• HTTP, limitations and order and meaning of
  messages
• Details of URLs
• Different Architectures
• Client Server
• Master Slave
• Distributed
• The Internet has nothing to do with Microoft at
  all

CCTM Course material developed by James King
(james.king_at_londonmet.ac.uk)