THE INTERNET & TCP/IP

THE INTERNET

• Some inventions cause a fundamental change in the way that we view our world, and other inventions accrete around it.
• The Internet is one of these.
• It began as a scheme for a decentralized network.
• It supports a suite of transfer protocols (TCP/IP).
• This led to hypertext, the browser, and many new ideas.
• Key ideas
  • Decentralized networking
  • TCP/IP
  • hypertext
  • World Wide Web

• Beginnings in 1964 at the RAND corporation
• The Problem: communicating among networked computers when part of the network was out of commission.
• Researcher Paul Baran described"distributed" computing, in a group of computers ("nodes") can be connected in such a way that damaged areas can be routed around.

Network Routing Configurations.
Source: Introduction to Distributed Communications Networks, Paul Baran

• The Question: How to control such a network? (Centralized control is vulnerable.)
• The Solution:
  • The network
    • would have no central authority; all nodes would be equal
    • would be designed from the beginning to operate with parts out of commission.
  • The transmitted messages would be divided into packets.
  • Each packet would be sent individually; every packet might take a different route to the destination.

• The Pentagon's Advanced Research Projects Agency funded a network using four high-speed supercomputers at ARPA contract sites. The purpose was to share computer resources.
• In 1969 the ARPANET had four nodes. By 1972 it had 37.
• The ARPA structure could accommodate many different platforms.
• The researchers started using the network for news and personal messages.
• ARPANET's original standard for communication was known as Network Control Protocol (NCP).
• NCP was replaced by a higher-level, more sophisticated standard known as TCP/IP.
• TCP/IP was being used by other networks to link to ARPANET, creating the Internet (Interconnected Networks).
• ARPANET went out of existence in 1989, replaced by the NSFNET.

TCP/IP

• The basic communication protocol of the internet is called TCP/IP (Transmission Control Protocol/Internet Protocol).
• TCP /IP is a two-layered program.
• The TCP layer
  • manages the assembling of a message or file into smaller packets that are transmitted over the internet
  • reassembles packets into the original message.
• The IP layer
  • handles the address part of each packet so that it gets to the right destination.
  • works like an envelope, with addressing at the beginning of your message. The addressing gives the network enough information to deliver the packet of data.
• packets are normally between 1 and 1500 characters long. This
  • reduces the chance of lost data
  • prevents any one user from monopolizing the network for a long period of time.
• IP addresses are 32-bit binary numbers (four bytes) , commonly written in a "dotted quad" form, such as 144.175.9.72
  • the first part tells the internet router what network the receiver is a part of.
  • The second part tells the network which computer should receive the packet.

TCP:

• breaks the data into smaller chunks (the packets)
  • numbers each packet, adding the number to the envelope (headers)
  • calculates a value called a checksum, adding the number to the envelope
    • the checksum allows the receiving TCP to detect error s in the packets.
  • passes the packets along to the IP layer
• collects received packets
  • puts the data back into the original order
  • checks to see if packets are missing
    • asks the sender to retransmit the missing packets if necessary
  • calculates what the checksum should be and compares it to the one received.
    • discards the packet and requests a retransmission if they don't match
• passes the data along to the application program using its services.

Domain Names

• we must have a way to identify locations on the Internet.
• numeric IP addresses accomplish this, but are difficult for humans to use.
• locations are be assigned names.
• names must be unique
• there must be a way to find a location (IP address) given a name
• there must be a way register new names.
• handled by the Domain Naming Scheme (DNS)
• similar to the way humans in the US are identified by both a fixed length number (SSN) and a name
• although human names are not necessarily unique, Internet names must be.

• when the Internet was still relatively small, dealing with names was simple.
• an agency called the Network Information Center (NIC) set up a registry.
• system administrators electronically submitted the desired hostname and corresponding IP address
• NIC added this to a file of names and addresses, called the hosts file
• the hostes file was distributed regularly to every machine on the network
• each computer used the hosts file to look up a name and find the corresponding IP address
• this is similar to looking someone's phone number up in a telephone book
• local networks still use this system. (all our Linux machines have a hosts file with the names of the others they are to communicate with.)

• as the Internet grew, this system became impractical
• this file got very large
• there were delays in getting name registereds
• it was very difficult to find names that weren't already used
• too much network time was spent distributing th is large file to every machine contained in it

• the solution was similar to the Internet itself
• a distributed, online system
• it is called the Domain Name System or DNS.
• the Domain Name System administers names by giving different groups responsibility for subsets of names
• Each level in this system is called a domain
• the different level domains in the complete name are separated by periods

•       cs.hood.edu
        nic.ddn.mil
        google.com

• the last part of the name (.edu, .mil, .gov, .com, etc.) is called a primary domain.
• agencies maintain the lists of the names assigned within that primary domain
• for example, the registries for the .com and .net domains are operated by VeriSign Inc.
• the group to which a name is assigned
  • can decide how to allocate the next level of names
  • maintains its own registry
• For example
  • names ending with the domain .hood.edu correspond to IP addresses beginning with 144.175.
  • a request by an off-campus browser or email server for cs.hood.edu must first find the necessary IP address for the packets
  • this is called a DNS Lookup
  • it first checks its local registry (called the DNS Server) to see if the answer is there.
  • if not, it must query the official registry for the primary domain (.edu) to find the first part of the address (144.175.)
  • the official registry also tells where to find the DNS server for .hood.edu
  • the lookup gets the next level of information from that registry. In the example, it is the rest of the IP address (.9.215).
  • if the lookup begins on the Hood network (another 144.175. machine), it will get the answer as soon as it queries the local DNS server.
• there could be another level of lookup
• you can obtain DNS information at http://www.internic.net/whois.html