Chapter1.md

Computer Networks and the internet

What is Internet

• Internet is a computer network that interconnects millions of computing devices throughout the world.
• The end devices are called hosts.
• End systems are connected by millions of communication links and packet switches.
• Many types of communication links exist. eg: copper wire, fiber optical, etc.
• The messages are divided into segments with some header information to be exchanged -> packets.
• Packet switch forwards packets arriving from one of its incoming comm link to one of its outgoing comm link.
• Eg: routers, link-layer switches
• Link-layer switches are used in access networks, routers in network core.
• Sequence of comm links and switches traversed by a packet is route / path.
• End systems access the internet via Internet Service Providers (ISPs)
• Lower and Higher level of ISP.
• Every internet component must communicate with each other using protocols.
• TCP/IP protocol.
• IETF creates standards for protocols conforming to what each protocol should be used for.
• These standards are called RFC -> Requests for Comments
• Internet can also be thought of as infrastructure that provides services to applications.
• End systems attached to internet provide an API that specifies how program can access the application, and how it will respond.

Network Edge

• End systems = Host: Those that sit at the edge of the internet (eg: laptops, computers, etc.)
• Hosts can be divided into clients and servers
• Access networks: Network that physically connects an end system to the first router.
• Home Access:
  • DSL - digital subscriber line - telephone
  • Cable Internet - cable tv
  • FTTH - fibe to the home
  • Satellite
  • Dial-up access - traditional phone lines
  • Local Area Network - LANs : wired and wireless (ethernet and wifi). end systems talk to access point (wireless) and which in turns talks to wired network.
  • Wide Area Network - WANs : 3g, 4g, 5g, etc.
• Physical Media is used to transmit bits from one component to another until source reaches destination.
• Physical media can be guided or unguided media:
  • twisted pair copper wire -> telephone, residential LAN access.
  • coaxial cable -> cable television
  • fiber optics -> high speed, overseas
  • terrestrial radio channels -> signals via em spectrum.
  • satellite radio channels -> geostationary and low-earth orbiting

Network Core

• packet switching:
  • Source breaks the messages into packets which are routed via communication links
  • If source sends at L bits/s and communication link can transmit at R bits/s then it takes L/R s to transmit all the bits via communication link
  • Store and forward transmission is adopted:
    • cause for transmission delay
    • packet switch must receive the entire bit before it can transmit.
    • a -> router -> b => L/R + L/R for the first packet. (total delay is 2L/R)
    • next bit starts at L/R (ie: after first packet has reached the router) => 3L/R second packet reaches the destination
    • and so on...
    • Total delay => N*L/R + (P - 1)*L/R for N links and P packets.
  • Queuing Delay and Packet Loss:
    • cause for queuing delay
    • each packet switch has multiple links attached to it.
    • packet switch has an output buffer (output queue) which stores packets that router is about to send into that link.
    • if link is busy, then packets are stored in queue. can lead to accumulation.
    • if buffer is full, might have to discard the packet => packet loss.
  • Forwarding tables and routing protocol:
    • every end system has an IP address.
    • router checks its forwarding table (hash table) to determine which is the next link it should send to based on packet's destination address.
    • forwarding tables are created using some routing protocols.
• circuit switching:
  • resources needed along a path (buffers, link) are reserved to provide communication between source and destination.
  • first host and desination establish a connection, send packets at a guaranteed constant rate as connection is already established (no waiting or delay)
  • connection gets 1/n of packet switching transmission speed as there are n links and each has dedicated connection reserved.
• Multiplexing in circuit networks:
  • FDM:
    • frequency division multiplexing
    • frequency of a spectrum is divided into different section / widths for different connections to use
  • TDM:
    • time division multiplexing
    • each connection is given a lot in the frame
    • every connection waits for it's turn in the order of the frame. (very small intervals, super fast)
• circuit vs packet switching:
  • packet switching is preferred because number of users > 10 that are simulataneously active are 0.1
  • packet switching allocates resources on demand, while circuit fixed resources beforehand
  • less expensive than circuit switching.

Netowrk of Networks

• Connecting end users and content providers to ISP + connecting ISPs with one another.
• Network 1: Interconnects all access ISPs with a single global transit ISP.
• Network 2: Multiple global ISPs avaiable for access ISPs to connect to. Global ISPs themselves are interconnected.
• Network 3: Multiple access ISPs connect to regional ISPs which connect to Global ISPs (Tier-1 ISPs)
• Network 4: PoP, multi-homing, peering and IXPs along with access, regional and global ISPs
• Network 5: Content providers creating their own network and connecting to different levels of upper ISPs
• PoP: Points of Presence => a group of routers in the provider's network where customer ISPs can connect to.
• Multi-homing: Customer ISP may choose to connect to two or more provider ISPs.
• Peering: Nearby ISPs at same level can directly connect to reach without exchaning any fee (free of cost)
• IXP: Internet Exchange Points => meeting point where multiple ISPs can peer together.

Types of delays

• Total Nodal Delay = Nodal Processing Delay + Queuing Delay + Transmission Delay + Propogation Delay
• Nodal processing delay: Examining packet's header information + sumcheck errors and deciding which link to send to. <= microseconds
• Queuing Delay: Waiting in the queue if the link is busy tranmitting other packets. >=0 to <=milliseconds
• Transmission Delay: Time taken to send the all bits in a packet to the link => L/R where is L: length of packets (number of bits), R: Tranmission Rate of switch. >=microseconds to <=milliseconds
• Propogation Delay: Time taken for all bits to reach from router a to router b => d/s => distance between two / speed of bits in the propogation medium. <=milliseconds
• Queuing delay depends on the traffic arrival. Traffic Intensity => La / R bits / s => a => average number of packets sent.
• If La / R > 1 infinite queuing delay, <= 1 queuing delay. = 0 => no delay. as the number of packets increase, queuing delay increasing drastically.
• Packet Loss: Queues have finite space. If traffic intensity approaches 1 then some packets may be dropped => lost as no space to accomodate the packets.
• If N-1 routers between source and destination => Delay end-to-end = N(Processing Delay + Transmission Delay + Propogation Delay) + K * Queueing Delay => K is some constant multiplier to consider traffic congestion.
• Traceroute is a program that traces the route a packet takes to reach from source to destination.
• It sends special packets that contain information on header that tells the routers that it's a traceroute program. The router sends back message to the source.
• The information received contains the router address, time taken, etc.

Throughput

• Rate at which destination receives the bits. -> (bits/s)
• Throughput depends on the transmission rate of the bottleneck link.
• If Rs is the rate between server and router, Rc is the rate between router and client => Throughput => min(Rc, Rs)
• If there are N links between server and client => Throughput => min(R1, R2, ..)
• In real world, rates of intermediate links are much higher than the rates of access networks.
• If x Routers are sending all information through a single network core of R transmission. Then throughput of core is considered as R / x which can affect the end to end throughput rate.

Protocol Layers

• Network designers organized different protocols into different layers -> protocol stack
• Internet Protocol Stack or TCP/IP Stack:
  • Application Layer:
    • network applications and application-layer protocols reside.
    • data => message
    • Eg: HTTP, SMTP, FTP, DNS
  • Transport Layer:
    • transports application layer messages between application endpoints by breaking them into segments
    • TCP (connection-oriented) and UDP(connection-less)
    • data => segments
  • Network Layer:
    • Takes the TCP / UDP segments and breaks it down into datagrams
    • Runs the IP protocol => How the fields must be defined, how to act on these fields
    • There are many routing protocols along with IP protocol.
    • data => datagrams
  • Link Layer:
    • Takes the datagrams and transfers it to the next node as frames.
    • some protocols ensure reliable delivery from one node to another.
    • each node can employ a different protocol along the route.
    • Eg: Ethernet, WiFi
    • data => frames
  • Physical Layer:
    • Transfers eaach individual bit of a frame from one node to another.
    • Protocols depends on the type of link and the medium.
    • data => bits
• Open Systems Interconnect Stack:
  • application, presentation, session, transport, network, data link, physical
  • presentation:
    • interpret the meaning of data exchanged
    • data compression, encryption, description
  • session:
    • delimiting and synchronization of data exchange
• When data is passed from one layer to another below, it's encapsulated with additional information

Network Security

• Malicious items - Malware - can enter and infect devices via the internet
• Compromised hosts - botnet - can be used by bad guys to control.
• Viruses - Malware that require human interaction
• Worms - Malware that don't require human interaction but can enter if the host is in vulnerable network
• DoS -> Denial of Service making service unusable for legitimate users.
• 3 types:
• vulnerability attack: crash the service
• bandwidth flooding: clog the target's access links with a lot of packets
• connection flooding: open a lot of half-open and full TCP connections, not letting other connections be established
• Bandwidth flooding can happen via DDoS attack -> Distributed Denial of Service when many compromised hosts (botnet) are controlled by the attacker to spam the service
• Packet sniffers are passive receivers that can be placed in the network, to sniff packets and decode sensitive information.
• IP spoofing is when the packet that is sent has a false IP address, masquerading as some user. => To prevent this, endpoint authentication must be established.

History of computer networks

• telephone network in 1969, used circuit switching
• first packet switched network was ARPAnet
• first host-to-host protocol -> Network control protocol (RFC 001)
• first email by Ray Tomlinson 1972
• tim berners lee developer html, http, web server and browser