History of the Internet

A Brief History of the Internet

• J.C.R. Licklider proposed the "Galactic Network" in 1962 with the vision of interconnected computers allowing quick data access.
• In 1962, Lawrence G. Roberts, an MIT researcher, connected computers in MA and CA using a low-speed dial-up line.
• ARPANET in 1969 demonstrated time-shared infrastructure and the need for packet switching.
• By the end of 1969, Roberts developed the ARPANET, connecting four nodes from UCLA, Stanford Research Institute, UCSB, and the University of Utah.
• In 1970, the Network Control Protocol (NCP) was introduced as an initial ARPANET host-to-host protocol, enabling network users to develop applications.
• Email was among the first applications, launching in 1972.
• In 1973, a DARPA team led by Bob Kahn introduced open-architecture networking, allowing independent design and development of individual networks.
• Kahn collaborated with Vint Cerf to develop the Transmission Control Protocol / Internet Protocol (TCP/IP), presenting it in 1973.
• TCP's first version divided functionalities into IP for addressing and packet forwarding, and TCP for flow control/lost packet recovery.
• The Domain Name System (DNS) was introduced by Paul Mockapetris at USC in 1983 to translate domain names to IP addresses using a scalable distributed mechanism.
• The World Wide Web (WWW) was introduced by Tim Berners-Lee and his team in 1990, becoming one of the most popular applications.

Internet Architecture Introduction

• A computer network is a complex system built on top of multiple components.
• Network protocols are structured into layers, providing structure to network architecture.
• Each layer offers different services.
• The functionalities in the network architecture are implemented by dividing the architectural model into layers.
• Analogy can be used to explain a layered architecture, for example, the airline system.
• Every layer works based on the service provided by the layer below it, and also it provides some service to the layer that is above.
• Layered network stack includes scalability, modularity and the flexibility to add or delete components which makes it overall easier for implementations that are cost effective.
• The Internet architecture follows a layered model, where every layer provides some service to the layer above.
• ISO proposed the seven-layered OSI model.
• The seven layers consists of the application layer, presentation layer, session layer, transport layer, network layer, data link layer, and physical layer.
• Separating functionalities into layers offers multiple advantages.
• Some disadvantages includes; some layers functionality depends on the information from other layers, which can violate the goal of layer separation.
• One layer may duplicate lower layer functionalities for example error recovery can occur in the lower and upper layers.
• There is some additional overhead that is caused by the abstraction between layers.
• The Internet architecture model has five layers
• The application, presentation, and session layers are merged into a single application layer.
• The interface between the application layer and transport layer uses sockets.
• Application developers design overall application functionality.

Application, Presentation, and Session Layers

• The application layer includes protocols like HTTP (web), SMTP (e-mail), and FTP (file transfer).
• The application layer offers depend on the implemented application.
• The presentation layer formats information between the layer below and the application layer.
• The session layer manages transport streams between end-user application processes, such as audio and video in teleconferencing.

Transport and Network Layer

• The transport layer handles end-to-end communication between hosts, using TCP and UDP.
• TCP offers a connection-oriented service, guaranteed delivery, flow control, and congestion control.
• UDP provides a connectionless, best-effort service without reliability, flow, or congestion control.
• The packet of information in the transport layer is called a segment.
• The network layer moves datagrams between Internet hosts.
• A source Internet host sends the segment with destination address to the network layer.
• The protocols in the network layer are the IP Protocol and the routing protocols which determine the routes datagrams can take between sources and destinations.
• The IP Protocol binds the internet together and must be run by all Internet hosts and devices.

Data Link Layer and Physical Layer

• The data link layer involves packets of information known as frames, with protocols like Ethernet, PPP, and WiFi.
• This layer moves frames from one node (host or router) to the next.
• The network layer routes datagrams across multiple routers, passing them to the data link layer for delivery to the next node.
• Services offerred depend on the data link layer protocol used over the link.
• The physical layer transfers bits within a frame between two nodes connected by a physical link, with protocols depending on the link and transmission medium, for example twisted-pair copper wire, coaxial cable, and single-mode fiber optics.

Layers Encapsulation

• Encapsulation and de-encapsulation are key concepts, showing the data's physical path from sending to receiving host.
• The transport layer appends header information (Ht) to the application layer's message (M), creating a segment.
• Network layer adds its header information (Hn) to the segment, forming a datagram.
• The link layer appends its header (Hl), creating a frame for physical transmission.
• At each layer, the message combines the payload (data from above) and the new header information from that level.
• At the receiving end, headers are stripped off in reverse order.
• Intermediate devices, like routers (layer-3) and switches (layer-2), also implement protocol stacks.
• Routers implement layers 1-3, while switches implement layers 1-2.
• The data leaves the sending host and is processed by the layer-2 switch via de-encapsulation and encapsulation.
• End-hosts implement all five layers, while intermediate devices do not, ensuring complexity and intelligence are at the network's edges.

The End to End Principle

• The end-to-end (e2e) principle shapes Internet architecture by suggesting application-level functions should not be built into the network core.
• Stated simply, the network core should be simple and minimal, while the end systems should carry the intelligence.
• A seminal paper on the subject argues that if a function can only be completely implemented with application endpoints, it cannot be a feature of the communications system itself.
• Not all applications need the same features and network functions to support them, building such functions in the network core is rarely necessary.
• This principle allows rapid Internet growth, with innovation at the network edge through services and applications, rather than the more-difficult-to-modify network core.
• Higher-level protocol layers are application-specific and more specific to an application, while lower-level layers organize network resources to achieve efficient application design goals.
• Data link protocols implement basic error correction because the physical medium is prone to interference and noise.
• Violations of the end-to-end principle typically refer to situations where it is not possible to implement functionality entirely at end hosts (NAT and firewalls).

Violations of the End-to-End Principle and NAT Boxes

• Firewalls and traffic filters violate the E2E.
• They monitor network traffic and prevent communication directly between end hosts.
• Network Address Translation (NAT) boxes use a bandaid to address the shortage of internet addresses.
• NAT-enabled home routers are an example of E2E violation.
• Internet Service Provider (ISP) typically assigns a single public IP address to the home router.
• NAT translation table rewrites source and destination IP addresses and ports.
• Hosts are not globally addressable/routable due to NAT boxes.
• Cannot initiate connections behind NAT without a NAT box.
• Workarounds: STUN (discovers NATs, public IP/port) and UDP hole punching (establishes UDP connections behind NATs).

The Hourglass Shape of Internet Architecture

• The Internet protocol stack resembles an hourglass shape.
• In the early 1990s, the network layer included multiple competing protocols, like Novell's IPX and X.25.
• Frequent innovations occur at the hourglass shape, and the waist protocols are hard to replace.
• The Evolutionary Architecture (EvoArch) model can study and explain layered architectures.
• This helps show lead how hierarchical structures lead to the hourglass shape.

Evolutionary Architecture Model

• Researchers suggest the EvoArch model, to answer previous questions on the internet design.
• Layers: Protocol stack is modeled as directed and acyclic network with L layers.
• Nodes: Each network protocol is represented as a node e.g. u.
• Edges: Dependencies between protocols are represented as directed edges.
• Node incoming edges: the service provided by a protocol w at the lower layer.
• Node substrates: S(u), nodes that are using the the services.
• Node outgoing edges: products of node P(u).
• Layer generality: layer is associated with a probability s(l).
• Layer generality decreases to higher layers.
• Node evolutionary value: value of protocol node, determined recursively.
• The value of the protocol depends on the values of the protocols that depend on it.
• Node competitors and competition threshold: the competitors of a node u, C(u), a fraction c of new node u products e.g. w.
• Node death rate: model uses a death rates.
• Protocols can cease.
• Competition occurs between nodes.
• Node basic birth process: a new node is assgned randomly.
• Toy Example
• 4 layers
• Generality probability s(I)
• EvoArch iterations perform the these steps.
• Introduce new nodes at layers.Connect new nodes through substrates.

Implications for the Internet Architecture and Future Internet Architecture

• The EvoArch helps explain the the success of TCP/IP, given that it existed when the telephone network was powerful.
• TCP/IP was not trying to compete with the telephone network.
• Used FTP, Email and Telnet.
• IPv4, TCP, and UDP provided a stable framework through which there is an ever-expanding set of protocols.
• The waist has large birth rate as it provides a stable waist through which there is great expansion of protcols.
• Suggests should design functionalily so waist is wide.
• What are the ramifications of the "hourglass shape of the internet"?

Architecture Redesign

• Internet is facing big challenges.
• The 4D group is redesigning Internet Architecture to optimize to optimize control and management to be direct at the network level.
• With functions spread across four components of the data.
• Suggests using control-plane acccountability improving controll-plane with original authenitation to detect misleading advertisements.

Interconnecting Host and Networks

• Repeaters and Hubs operate on physical layer and are simple.
• Bridges and Layer2-Switches can enable communication between host using Media Access Control (MAC) .
• Routers and Layer3-Switches operate at layer 3.
• Learning Bridges- device with multiple input/ouputs the transfers frames from one input or mutiple the are populated through populating a forwading table.
• Looping Problem in Bridges and the Spanning Tree Algorithm: Excludes links that lead to loops by running the spanning tree alograthism where select links should be used for forwading to eliminate loops, the algo uses forwading tables.
• To prevent attacks on routing. the propose work suggests improve control-plane e accountibilty providing originating authentication ensuring the network that that appears authentic.