8. Application Layer

Introduction

• responsibilities

  • uses the functionalities/proptocols provided by Transport Layer & Network Layer (TCP/IP), and provides specialized services to the user like mail transfer, file transfer, hypertext transfer and thousands others
  • types of protocols
    • standard(like ftp, http)
    • non standard (anybody can create and use privately)
  • two machines on internet can communicate with each other with the help of application layer using one of two paradigms
    • client server paradigm
    • peer to peer paradigm

• client server paradigm

  • server

    • an application program that provides services

  • client

    • the application program that uses server's services

  • server process

    • server process runs continuously waiting for client process, to make connection and ask for service
    • server processes provide a specific type of service

  • client process

    • the client process is started when client needs to recieve service
    • the client process can request service from any of the server provessess

  • examples

    • WWW, e-mail etc.

  • problems with client server paradigm

    • the load of communication is more on the server process, so the server should be powerfull computer
    • a service provider must be willing to bear cost of setting up such costly server, so they will need income from it, which is against free sharing ideology of the internet

• peer to peer paraidigm (p2p)

  • the computers on the internet
    • can provide service at some time
    • or can use the service provided by other computers on internet
    • or do both of these together
  • example
    • BitTorrent

Client Server Programming

• Application Programming Interface (API)

  • what

    • the OS manufacturer should build first four layers of TCP/IP suite in OS
    • and it should also provide easy set of instructions to use the services of TCP/IP layers
      • like to open connection, send/recieve data from other end, close connection
    • thses set of instruction are known as API

  • types of API's designed for communication

    • socket interface
    • Transport Layer Interface (TLI)
    • STREAM

  • Socket Interface

    • what?
      • a set of instructions used for communication b/w OS and application layer
    • socket?
      • socket is supposed to be like a sink, that can be accessed like a file, by any programming language
      • so, in a client/server paradigm, communication takes place b/w two sockets
        • so computer think that it is just communicating with socket
    • how?
      • if we create sockets at each
      • and define source and destination socket address correctly
      • we can use available instructions to send and recieve data
        • how? we don't need to know,TCP/Ip takes care of it
    • socket address?

      • the one, that underlying TCP/IP protocol needs to reach application at the other end..

      • yes IP address and port number, is together known as socket address, used by socket

        • 

      • finding socket address

        • server site
          • the server knows its local IP and local port is provided by OS, unless it is a standard process
          • the server knows the IP and port of client when it connects with it
        • client site
          • the client knows its IP, while OS provides port no. to it (one of non standard)
          • the client generally knows port no. of server as it is one of standard. To knon IP, it types just the identifiers(URLs), and the DNS gives the IP to it

Domain Name System (DNS)

• what
  • it acts like a phone book, that matches b/w domain name (url) and IP address
  • we can have a dynamic IP, as we can change it to whatever we want, but domain name stays same
• server
  • it is a heirarchial server
  • root server (it has info about domain servers)(various root servers are there around the world, a DNS client contacts the one closer to it)
    • generic domain servers (they have info about the sub domain servers)
      • subdomain servers (they have informatino about the domain name)
• process of aquiring IP by DNS
  • the machine uses the DNS client found on the computer
    • although sometimes the machine first asks DNS resolver for the IP, which might have cached it
    • if not, then the resolver goes and fetches the IP, and caches it and returns to the machine
  • it sends a request to the DNS server whose address is already in the computer
  • the DNS server finds the IP for the given domain name and returns it
  • now the machine gets the IP from the DNS client, and can use socket to communicate with another application layer

Application Layer Standard Protocols

• 0 to 1023 port numbers given to them

• examples

  • Echo (7, TCP/UDP)
    • to check roundtrip time from a server, to check if we are connected to network
  • FTP (20/21, TCP)
    • file transfer protocol, needs reliable protocol
    • 20 used for data, 21 used for control
  • SSH (22, TCP)
    • secure shell
    • a tunnele is made
  • Telnet (23, TCP)
    • used for remote login
  • SMTP (25, TCP)
    • simple mail transfer protocol
  • DNS (53, UDP)
    • domain name system
    • sends/recieves messages so uses UDP
  • DHCP (67/68, UDP)
    • gets IP address by mac address
  • TFTP (69, UDP)
    • trivial file transfer protocol
    • file send without connection
  • HTTP (80, TCP)
    • hypertext transfer protocol
  • POP (110, TCP)
    • to pop messages from email
  • NTP (123, UDP)
    • network time protocol
    • to sync clock/time
  • HTTPS(443, TCP)
    • http secured
    • encrypts packets
  • RIP (520, UDP)
    • routing information protocol
    • DVR (distance vector routing protocol)

• HTTP

  • port no 80
  • to fetch web pages
  • it is not reliable but uses TCP for reliable
  • inband protocol (commands and data go to same port)
  • stateless (http doesn't store state)
  • HTTP 1.0
    • non persistent
    • connection doesn't persist if we close site
    • doesn't make connection, so connection making overhead is more
  • HTTP 1.1
    • persistent
    • connection persists, if we open site again, we will be there
    • makes connection and maintains, so connection making overhead is less
  • commands
    • head, get, post, put, delete, connect
  • HTTPS is more secure than http
  • works with http request and response
  • methods
    • get (requests document from server)
    • head (requests info about document)
    • post (send some info to server)
    • put (send document from server to client)
    • trace (echos incoming request)
    • option (inquires about options)

• FTP

  • port no. 20(Data) & 21(control)
  • data connection is non persistent
  • control connection is persistent
  • not inband
  • reliable (can't lose data, uses TCP)
  • stateful (maintains state at server end)

• SMTP & POP

  • FTP is synchrounous(client & server need to be connected) but SMTP & POP is both synchronous and asynchronous
  • SMTP port no. 25 for pushing mail
  • POP3 works with two ports
    • port 110 (default, non encrypted)
    • port 995 (secure connection)
  • MIME (multipurpose internet mail extensions)
    • for images, videos etc

• WWW (world wide web)

  • so ya, this is like the web
  • it is repository of information in which documents called webpages are stored
  • this repository is
    • distributed
      • each web server in the world can add it's own web page and announce it to all internet users
    • linked
      • clicking on various hypertext takes us on another page
  • it is a distributed client server service, in which client uses web browser to access services, using URL (uniform resource locator)
    • the url is converted to ip with help of DNS and appropriate connection is made
    • and voila, you are surfing on the web