Computer Networks and the Internet

Table of Contents

    1 What Is the Internet?
    2 The Network Edge
    3 The Network Core
    4 Protocol Layers and Their Service Models

The Internet is the worldwide computer network, where billions of desktop and mobile devices connect to each others via packet switches and communication links

What are you going to see in this section?

The basic hardware and software components that constitute a network. At the network edges there are: end systems and network applications; at the network core there are: 1) links and switches transporting data, 2) access networks connecting end systems to the network core, 3) interconnection between IPSs to build the Internet as a network of network. An introduction to architectural principles of computer networking such as protocol layering and service models

1 What Is the Internet?

We use the Internet to discuss of computer networks and their protocols.
There are two ways to say what is the Internet: 1) describe the basic hardware and software components that make up the Internet, 2) describe the networking infrastructure that provides services to distributed applications

1.1 Explain what the Internet is by describing its components

AT NETWORK’S EDGES: HOSTS OR END SYSTEMS

The Internet is a computer network that connect billions of devices across the world. Desktop computers, workstations, servers, mobile devices (such as smartphones and tablets) IoT devices (such as: TVs, game consoles, home appliances, wearables, surveillance systems, cars and more) are connected to the Internet. In the Internet dialect, these devices are named hosts or end systems.

Figure: some pieces of the Internet

AT NETWORK CORE: LINKS AND SWITCHES

Communication links and packet switches transport data: end system are connected by a mesh of links and packet switches

Communication Links

Communication links connects two or more devices and make it possible data transmission.
There are many different types of communication links, such as: radio waves (for example IEEE 802.11 Wi-Fi and mobile networks), electrical impulses (for example RJ-45 Ethernet), light optics.
Each type of communication link is made up of a different kind of physical medium/material: such as radio frequency spectrum, copper wire, optical fiber.
Each type of link transmits data at different transmission rate or bandwidth, which is measured in bits/second

Packet switches: router and link-layer switch

If an end system has to send data to another end system, the sending system segments the data into smaller segments, includes a header in each segment and send these segments through the network to the destination end system, where they are reassembled into the original data. Those segments of data are also known as packets.
A packet switch takes a packet arriving on one of its incoming communication links and forward the packet on one of its outgoing communication links.
There are two types of packet switches: router and link-layer switches; the link-layer switches are typically used in access networks, whereas the routers are typically used in the network core.
The route or path (through the network) is the sequence of links and switches that a packet must traverse from the sending end system to the receiving end system.
It is possible to make an analogy between transportation networks and packet-switched networks. Packet-switched networks (which are made up of mesh of communication links and packet switches) transport packets; in the same way transport road networks (which are made up of a mesh of highways and intersections) transport vehicles. For example: a factory move a cargo from a production site building to a warehouse building using truck vehicles; in same way, an end system send data to another end system using data packets. So, the packets are analogous to the vehicles, communications links are similar to highways, packets switches are similar to intersections and end systems are analogous to buildings.

AT NETWORK CORE: ACCESS NETWORK

An access network are the physical infrastructure and technologies that connect end users to the network core. Internet Service Providers (ISPs) are the companies that provide Internet connectivity to end systems, through the access networks.
Types of ISPs are: ISPs that provide Internet access to residential customers (such as cable or telephone companies), ISPs that provide Internet to corporate employees, ISPs that provide Internet access to Universities, IPSs that provide WiFi access in coffee shops, airports, hotels, etc. Cellular data ISPs, which provide a mobile access to smartphones.
ISPs provide different types of network accesses to end systems: 1) broadband access, such as: cable modem, DSL and fiber to the home 2) high-speed local area network access, 3) mobile wireless access. Note: ISPs also provide Internet access to content provider’s servers.

AT NETWORK CORE: ISPs INTERCONNECTION

Interconnection between IPSs builds the Internet as a network of network
The Internet is about connecting end systems to each others, so the IPSs have to be interconnected;
ISPs are organized in tiers: lower tier IPSs provide access to end systems and are connected to national or international higher-tier ISPs, higher-tier ISPs are interconnected directly to each others via fiber-optic links.
Both lower and higher tier ISPs run the IP and DNS protocols.

Protocols

End systems and packet switches use protocols to control the transmission of information in the Internet.
The Transmission Control Protocol (TCP) and the Internet Protocol (IP) are the two most important Internet protocols and are collectively known as TCP/IP.
Internet protocols are described by standards, so that people can create systems that interoperate.
The Internet Engineering Task Force (IETF) publishes Requests For Comment (RFCs) which may later be adopted as Internet Standards.

1.2 Explain what the Internet is by describing its services

INTERNET APPLICATIONS

Internet applications (also known as distributed applications) are programs that run on end systems; packet switches allow the exchange of data among end systems, where applications are the source or destination of data.
Internet applications include email, web browsers, messaging, music and video streaming, video conferencing, multi-player games...

INTERNET AS PLATFORM FOR APPLICATIONS, THE SOCKET INTERFACE

The Internet can also be described as an infrastructure that provides services to applications;
suppose a developer has an idea for a new Internet application and therefore he writes its program. This program, which run on an end system, needs to tell Internet to deliver data to another program running on another end system.
A program that run on an end systems, connected to the Internet, has access to the Internet socket interface. The Internet socket interface defines a set of rules that a program that needs to send data to a destination program has to follow to tell the Internet to deliver the data.
We can employ an analogy: if Alice wants to send a letter to Bob, the postal service has a set of rules that Alice has to follow if she wants to send its letter. The postal service has its own “postal service interface”.

1.3 What Is a Protocol?

A HUMAN ASKING FOR TIME ANALOGY

Let's make a human analogy, to understand computer network protocols. Consider what a person does to ask someone for the time of day: to start the communication, one first greets with a “Hi” and, if he gets a “Hi” message as response, proceeds and ask for the time. Another response or no response at all, indicates that the person is not able or does not want to communicate. Messages exchanged and actions taken play a role in the human protocol: if a person does not understand English or does not understand the concept of time, the protocol does not interoperate and the two humans cannot work together. Also in networking, it takes to entities that use the same protocol to complete a task.

NETWORK PROTOCOL

A network protocol is different from a human protocol in that the entities exchanging messages and taking actions are not persons, but software/hardware components of a network capable device.
Each activity that involves two remote entities that communicate is controlled by a protocol.
Protocol examples: two computers implement in hardware the physical layer protocols that control the flow of bits on the wires between their network interface cards; routers implement protocols to determine the path of a packet from source to destination.

NETWORK PROTOCOL EXAMPLE: ASKING FOR A WEB PAGE

Example of network protocol: making a request to a web server.
When you type an URL in the web browser, your computer will send a TCP connection request message to the web server and wait for the reply, the web server receives your connection request and returns a TCP connection response message. If the computer gets an OK, it will send a HTTP request message, which contains the name of the web page to fetch; finally, the web server returns a HTTP response message containing a file of the web page to your computer.

DEFINITION OF PROTOCOL

A protocol defines the format and the order of messages exchanged between two or more communicating entities, as well as the actions taken on the transmission and/or receipt of a message or other event.

Protocols are pervasive in computer networks, as they allow to accomplish many communication tasks

2 The Network Edge

END SYSTEMS

As we already said, the Internet dialect's term “end systems” is used to indicate devices connected to the Internet; they are said end systems because they are on the edge of the Internet.
Internet end systems include: desktop computers, servers, mobile devices and IoT devices.

NETWORK HOSTS AND NETWORK NODES

End systems are also said hosts, because they host Internet applications such a web browser/server, an email client/server. Hosts are classified into two groups: clients and servers. Clients are likely to be desktops, laptops, smartphones and so on, whereas servers are high-performance machines which are capable of storing and distribute web pages, email, videos, etc. to clients. Now a day’s servers reside in big data centers; for example, Google has 30 data centers in the four continents.

A network node is any device participating in a network, whereas a host is a node that runs user applications; every network host is a node, but not every network node is a host. Network infrastructure hardware, such as modems and packet switches are considered to be node but not hosts, because they do not have application level functionalities.

2.1 Access Networks

WHAT IS AN ACCESS NETWORK?

An access network connects end systems to the edge router, which in turn connects to the network core and to any other remote end system

Home Access: DSL, Cable, FTTH, and 5G Fixed Wireless

Most of the houses in the western world has an internet access: the most common types of home access networks are digital subscriber line (DSL) and cable.

DSL ACCESS

A house gets a DSL internet access from the same telephone company (telco) which provides its phone access, so the telco is also the ISP.

A DSL Internet access uses the telephone company’s pre-existing telephone line. The customer’s DSL modem exchanges data with a digital subscriber line access multiplexer (DSLAM) at the telco local central office; the DSL modem translates digital data to high-frequency tones to be transmitted on telephone wires to the central office, where the DSLAM translate them back to digital format.

The telephone line carries data and telephone signals at same time; each signal is encoded in a different frequency band: 1) a common two-way telephone channel in the [0, 4 kHz] frequency band, 2) a upstream channel in the [4 kHz, 50 kHz] frequency band, 3) a downstream channel in the [50kHz, 1 MHz] frequency band. In this way, Internet and telephone connections share the same DSL link, like it there were three links not one. At the customer's house, a splitter separates the data and phone signals arriving to the house; at the telco central office, the DSLAM separates the data and phone signals. Many hundreds of houses connect to a single DSLAM.

CABLE ACCESS

A cable Internet access uses of the cable television company’s pre-existing cable television infrastructure. A house gets a cable Internet access from the same company that provides its cable television.

A cable television headend receives the television signal; a fiber optics connect the cable headend to neighborhood-level junctions; from these junctions the coaxial cable is then used to reach individual houses. Because this system employs both fiber and coaxial cable, it is often named hybrid fiber-coaxial (HFC).
The customer’s cable modem exchanges data with a cable modem termination system (CMTS) located at the Cable headend; the cable modem converts digital data to analog data to be transmitted on coaxial cables to the cable headend where the CMTS translate them back to digital format.
Cable modems divide the signal into two channels: a downstream and a upstream channel; as with DSL, cable access is asymmetric.

FTTH

The fiber to the home (FTTH) technology employs an optical fiber line from the telco’s local central office to the home. FTTH can provide Internet access at gigabits per second rates, which is higher speed than DSL and cable networks.

There are many different technologies for the optical distribution from the telco’s central office to the houses: the simplest one is named direct fiber, with one fiber from the telco’s central office to each house, but usually each fiber leaving the central office is shared by many houses. Indeed, when the fiber is close to customers homes, it is split into individual customer specific fibers. There two different network architectures that perform this splitting: the active optical networks (AONs) and passive optical networks (PONs).

Figure FTTH Internet access

The PON distribution architecture is illustrated in figure 1.7: each home has an optical network terminator (ONT), all these ONTs are connected with a single fiber to a neighborhood splitter; the splitter combines a number of homes into a single shared optical fiber, which connects to optical line terminator (OLT) at the telco’s local central office. The OLT supplies convertion between optical and electrical signals and connect to Internet through a telco router. At home, users connects a home wireless router to the ONT (using an Ethernet cable) to access the Internet through this router. The ONT converts light signals from the OLT’s fiber optic line into electronic signals for use by your router, working as a modem.
Note: the Optical Line Terminator serves as ISP’s endpoint, whereas an optical network terminator constitutes the endpoint of the PON at subscriber's home.

5G

5G fixed wireless is a high-speed residential access, which can do without installing expensive cables from the telco’s local central office to the home.
5G technology can send data wirelessly from the provider’s base station to a residential modem which holds a 5G sim.

Ethernet and WiFi

LANs as access networks

Universities and offices set up LANs to connect end systems to the edge router.
Ethernet and WiFi are the two most common technologies in use for connecting computers in local area networks.

WIRED (ETHERNET) LANS

Ethernet LAN users employ a twisted-pair copper wire to connect to an Ethernet switch, the Ethernet switch is then connected to the Internet. Ethernet has transmission rate from 100 Mbps to 10 Gbps to the Ethernet switch.

Figure: Ethernet Internet access

WIRELESS (Wi-Fi) LANS

Wireless LAN users exchange packets with an access point which is connected to the Internet. WiFi access in now present almost everywhere: universities, offices, cafes, airports, home and even in trains and planes.
Wireless LAN are bases on the IEEE 802.11 technology, also known as WiFi. A wireless LAN user have to be within a few tens of meters of the access point. WiFi provides transmission rate up to 100 Mbps.

LANs at HOME and ENTERPRISE SETTINGS

At the beginning Ethernet and WiFi access networks were deployed in enterprise settings, but now are also common in home networks. Many homes use both a broadband residential access and a wireless LAN technologies to set up their networks.

The figure below shows a home network with mobile devices, many internet connected home appliances, wired desktop computers, the wireless access point or base station that communicates with the wireless devices in the home, a home router that connect the wireless access point to the Internet.

Figure: A typical home network

Wide-Area Wireless Access: 3G, LTE 4G and 5G

WWANs

Mobile users make use of their smartphone to run social media apps, to consume video and audio streaming, to exchange messages, to make mobile payments and much more. These devices use of the provider's cellular infrastructure both for mobile telephony and for exchanging packets.

Telecommunications companies implement mobile telecommunication cellular network technologies such as 2G, 3G, 4G LTE and 5G to transfer data. These technologies are sometimes referred to as Mobile Broadband. The fourth generation (4G) wireless provides download speed of up to 60 Mbps. A cellular network user has to be within a few tens of kilometers of the base station.

2.2 Physical Transmission Media

Each network access technology uses its own physical media, for example DSL and Ethernet uses copper wire and the mobile access networks uses the radio spectrum.
Let’s consider the journey of bit traveling from one end system to another through a series of links and routers. The source end system transmit the bit and soon afterwards the first router in the series receives the bit; the first router then transmits the bit and soon afterwards the second router receives the bit and so on. The bit passes through a series of transmitter-receiver pairs. For each pair, the bit is sent by electromagnetic waves or optical pulses, which both propagate through a physical medium.

Physical media can be categorized in guided media (waves propagating in wires and cable) and unguided media (waves propagating in space or atmosphere, such as LAN and satellite). Examples of physical media are: twisted-pair copper wire, coaxial cable, fiber-optic cable, terrestrial radio spectrum and satellite radio spectrum.

Twisted-Pair Copper Wire

It is the most common guided transmission medium. It has been used by telephone networks for a hundred year: the wired connections from the telephone device to local telephone switch uses twisted-pair copper wire.

Twisted pair is commonly used for residential Internet access in dial-up modem technology and DSL technology. Modern twisted-pair technology is still the preferred solution (over fiber-optic) for high-speed LAN networking.

Twisted pair consists of two insulated copper wires arranged in a regular spiral pattern; a pair is wrapped in a protective shield; a cable may contain a number of pairs. A wire pair constitutes a single communication link. Unshielded twisted pair (UTP) is a cheap wire commonly used to set up LAN networks within a building.

Coaxial Cable

Coaxial cable is used in cable television system and cable Internet access as guided shared medium.
Coaxial cable consists of two copper conductors, like twisted pair. Unlike twisted pair, the two conductors are concentric not parallel.

Fiber Optic Cable

A fiber-optic cable, aka optical-fiber cable, contains one or more optical fibers.
An optical fiber transmits pulses of light, rather than electrical signals
A single optical fiber can support very high bit rates, up to tens or hundreds of gigabits per second.
Fiber optics are not affected by electromagnetic interference and have low signal attenuation; for this reason they are the preferred guided transmission media for long distance.
Optical fibers are used for long distance telephone networks, Internet backbone and overseas links.
Anyway, the high cost of optical devices has slowed down their deployment for short distance transport, for example in LAN and home access networks.

Terrestrial Radio Channels

Electromagnetic waves can carry signals in the electromagnetic spectrum. Radio waves are electromagnetic waves of frequency between 3 Hz and 300 GHz. A radio channel is a specific radio frequency used to convey a signal.

Radio channels do not require physical wires, can penetrate wall, provide connectivity to mobile users and travel long distances. On the other side, radio channels are subject to path loss, fading (due to obstructing objects) and interference (due to other transmissions)

Terrestrial radio channels can be classified in three groups: radio channels operating over very short distance, one or two meters, such as wireless headset and keyboard; radio channels operating in local areas, ten to a few hundreds meters, such as the wireless LAN technology; radio channels operating in the wide areas, tens of kilometers, such as the cellular access technology

Satellite Radio Channels

A communication satellite connects two or more base earth stations aka ground stations. A earth station uses parabolic antenna to transmit or receive information by radio microwaves to or from a communication satellite.
There are two types of communication satellites: geostationary (GEO) satellites and low-earth orbiting (LEO) satellites.

• GEO satellites are placed in orbit at 36,000 kilometers above Earth’s surface and they always stay above the same place on Earth, whereas LEO satellites are placed at 160 to 2,000 kilometers above Earth surface and they rotate around the Earth, like the Moon does.

• GEO satellites are used in areas without DSL, cable or fiber optics. LEO satellites may be used for Internet access in the future.

3 The Network Core

The network core is made up of a mesh of packet switches and links connecting the Internet end’s systems. See figure below.
This section examines switching and routing in computer networks.

Figure: The network core

3.1 Packet Switching

PACKETS AND PACKET SWITCHES

Network applications, running on end systems, exchange messages with each others.
A source end system, which want to send a message, breaks the message in smaller chunks of data also known as packets. Each packet travels on a path between the source end system and the destination end system, the path is made up of communication links and packet switches.
A packet switch may be of two types: router and link-layer switch. A packet switch has many input and output ports where links are attached to. A packet switch switches an incoming packet onto an outgoing link.
Suppose that a communication link has a transmission rate of R bits/sec, if an end system or a switch is sending a packet containing L bits on the link, the time to transmit the packet is L/R seconds.

Store-and-Forward Transmission

Packet switches uses a technique named store-and-forward transmission. Store-and-forward transmission means that the packet switch first receives all the bits of a the packet and then it may begin to transmit the first bit of the packet on the outbound link.

To understand store-and-forward transmission, consider a simple network (see the figure below) consisting of two end systems connected by one router, where the router has only two links.
Suppose that the source wants to transmit three packets, each of L bits; at the snapshot of time in figure, the source has transmitted some bits of the packet 1 and these bits are already arrived at the router, the router cannot transmit these bits, it has to store them in a buffer. Only when all the bits of the packet have been received, the router may begin to forward the packet onto the outbound link.

Time to trasmit one packet

Let’s calculate the amount of time that elapses from when the source begin to send the first packet until the destination has received the entire packet.
The source start transmission at time 0, at time L/R seconds the source has finished to transmit and the router has received and stored the whole packet. At time L/R, the router can begin to forward the packet. At time 2L/R, the router has transmitted the entire packet, which has been received by the destination. Therefore, the total delay is 2L/R.

delay = 2 (L/R)

Delay of a packet across a path of 2 links and 1 router Link rate R, L bits for packet.

Figure: Store-and-forward packet switching

Time to trasmit three packets

Let's consider the transmission of three packets and calculate the amount of time that it takes from when the source starts to send the first packet until the destination has received all the three packets. At time L/R the router start forwarding the first packet and the source start sending the second packet. At time 2L/R, the destination has received the first packet and the router has received the second packet. At time 3L/R, the destination has received the first two packets and the router has received the third packet. Finally, at the time 4L/R, the destination has received all the three packets.

delay = 4 (L/R)

Delay of 3 packets across a path of 2 links and 1 router

Time to trasmit one packet across N links

If we consider the case of a packet that travels from a source to a destination across a path consisting of N links each of rate R. By using the same logic as before, we can say that the end-to-end delay is

delay = N (L/R)

Delay of one packet across a path of N links. Link rate R, L bits for packet.

Queuing Delays and Packet Loss

A packet switch has multiple links attached and uses a output buffer (aka output queue) for each attached link to store packets the router is about to send on that link.
If a packet arrives at the router which is busy with the transmission of another one, the arriving packet has to wait in the output buffer. Therefore, a packet may experience two delays: the store-and-forward delay and the output buffer queuing delay; these two delays are variable and depend on the congestion in the network.
Due to the fact that the buffer is finite, an arriving packet may find the buffer full with other packets waiting to be forwarded; in this case, there will be a packet loss: either the arriving packet or the queued packet is dropped.

Forwarding Tables and Routing Protocols

Packet switching and forwarding tables

The job of a router is forwarding packets to another router; to accomplish its job, the router switches a packet from an input link to an output link.
Each type of computer network does packet forwarding its way. In the Internet, a router determine what is the appropriate link it should forward the packet onto based on the destination address the packet carries. In fact, in the Internet, every end system has an IP address. When a source end system want to send a packet to a destination end system, the source adds the IP address of the destination system in the packet’s header.
Each router has a forwarding table that maps portions of IP address to its outbound links; the IP address has a hierarchical structure like the geographical address. When a packet arrives at the router, the router examines the destination address and uses the IP or part of the IP to search in the forwarding table and find the appropriate outbound link to direct the packet.
Note: the act of forwarding packets to another router is referred to as packet switching

Forwarding tables and routing protocols

A router creates a forwarding table that maps IP address to its outbound links and uses this table to do packet forwarding. How do forwarding tables are set? The Internet has routing protocols that are automatically used to configure the forwarding tables. For example, a routing protocol determines the shortest path from each router to each destination and uses this path to configure the router's forwarding tables.

3.2 Circuit Switching

Packet switching and circuit switching are two alternative ways to communicate on a network of links and switches; this section covers circuit-switched networks.

Packet-switched networks VS Circuit-switched networks

In circuit-switched networks, to set up the communication the required resources is reserved by the end systems for the entire duration of the communication session; whereas in packet-switched networks, during the communication session, application messages use the resources on demand and they may have to wait for access to resources.
As an analogy, consider having a dental treatment or having a urgent care treatment. For the dental treatment you have to book an appointment, but as soon as you arrive at the dentist the doctor (reserved resource) takes care of the problem. For the urgent care treatment, you rush to the hospital without an appointment, but when you arrive you may have to wait for your turn to see a doctor (not reserved resource).

Circuit-switched network example

Traditional telephone network, aka Public Switched Telephone Networks (PSTN), let users make landline telephone calls and are an example of circuit-switched data transfer. What happens when a person wants to send information to another person on a telephone network? Before the sender is able to send the information, the network has to establish a dedicated end-to-end connection between the sender and the receiver.

What is a circuit-switched connection?

The circuit-switched connection is a genuine connection in which a physical path is obtained and dedicated to a single connection.
What is important about circuit switching is that the physical path stays alive and engages the switches and the links for the duration of the connection, with the switches on the path maintaining the connection's state throughout the communication session. The connection can garantees a constant transmission rate on the network’s link (representing a fraction of the link’s transmission capacity). In the telephony jargon, this connection is called a circuit.

The figure below shows a circuit-switched network. In this network, the four circuit switches are interconnected by four links. Each link has four circuits, and therefore each link can support four simultaneous connections; the hosts are connected to one of the switches.
When two hosts want to communicate, the network must first establish a dedicated end-to-end connection between the two hosts. So, the network reserves one circuit on each of the two links. Since each link has four circuits, the connection gets one fourth of the link’s total transmission capacity. If a link between two switches has a transmission rate of 1 Mbps, then each end-to-end circuit-switch connection gets 250 kbps of dedicated transmission rate.

Figure: a circuit-switched network with four switches and four links

On the other side, if a host want to send a packet to another host on a packet-switched network, for example Internet: like circuit-switching, the packet is transmitted on a series of communication links; unlike circuit-switching, the network does not reserve any link resource for the packet transmission. If one link is congested, then the packet has to wait in a buffer of a transmission link.

Multiplexing in Circuit-Switched Networks

What is Multiplexing?

In telecommunications and computer networking, multiplexing or muxing is a way to combine multiple signals into a single composite signal that is conveyed over a shared communication medium, multiplexing can be applied to both analog and digital signals.

How to implement a circuit in a link?

As we already said a circuit-switched connection is also said a circuit.
A circuit is implemented with either frequency-division multiplexing (FDM) or time-division multiplexing (TDM). With FDM, the frequency spectrum of a link is divided in frequency bands and each band is dedicated to a single connection for the duration of the connection. With TDM, the time for which the link is used is divided into time frames and each frame is divided into a fixed number of time slots; when a connection is established, each time slot in every frame is dedicated to this connection.

Examples of applications of FDM and TDM

Examples of applications of FDM are telephone systems, radio and television broadcasting; in telephones the frequency band typically has a width (or bandwidth) of 4 kHz; in FM radio the frequency spectrum between 88 MHz and 108 MHz is divided in 100 frequecy bands, aka channels, each band is 200 kHz wide and each station is allocated at a specific channel.
Examples of applications of TDM are in the telecommunications field, such as in telephone systems, optical networks, and satellite communications.

Figure 1.14 demonstrates FDM and TDM multiplexing strategies for a specific link which supports up to four circuits. For FDM, the frequency domain is divided in four bands. For TDM, the time domain is divided into frames, with four time slots in each frame; a cicuit is assigned the same time slot in the sequence of time frames.

Packet Switching Versus Circuit Switching

Let’s compare circuit switching and packet switching.

• Cirtcuit switching drawbacks 1) Circuit switching it may result in a waste of resources, because the dedicated circuit is idle during the time a user does not send data. For example, if a person is making a phone call and he stops talking for some time the network resources (frequency bands in FDM or time slots in TDM) are idle and cannot be used by the other active connections. 2) establishing a circuit is difficult and time wasting.
• Packet Switching advantage: it more efficient than circuit switching as it provides better sharing of transmission capacity than circuit switching
• Packet Switching drawback: it is not suitable for real-time services, for example telephone calls and video conference calls, because of its variable queuing delays.

Examples: packet switching is more efficient than circuit switching

• Suppose that a 1 Mbps link is shared by multiple users. Suppose that each user alternates between period of activity generating data at 100 kbps and period of inactivity when it generates no data. Suppose that a user is active only 10 percent of the time. With circuit switching, 100 kbps is reserved for each user all the times; therefore, the circuit switched link can only support 10 simultaneous users. With packet switching, the probability that a user is active is 0.1 or 10%, and with 30 users the probability that there are 11 or more simultaneously active users is still very low. So, packet switching provides the same performance as circuit switching, but allows more than three times the number of users.

• Suppose that there are 10 users and that one user generates a burst of packets, while the other users remain not active. With TDM circuit switching, there are ten slots per frame and only one slot is active, so the active user takes some time to transmit all its data; whereas with packet switching, the active user can send data at the full link rate.

In conclusion, packet switching and circuit switching are both used in telecommunication networworks, but many networks are being converted to packet-switched, also circuit-switched telephone networks are moving to packet switching.

3.3 A Network of Networks

End systems (mobile, desktop and servers) connect to the Internet using an access ISP.
The ISP provides end systems with wired or wireless connectivity using various access technologies: DSL, cable, optical fiber, WiFi and cellular.
To connect these end systems to the Internet, it does not suffice that end systems connect to an access ISP, but also access ISPs themselves have to be interconnected. This creates a network of network.

ISPs establish the worldwide connectivity and are organized in a hierarchy with several tiers

• tier-1 IPSs are at the top level,
  they consist of big telecommunication companies that exchange traffic with each other through a high speed network of fiber-optic links and routers. There are approximately a dozen tier-1 ISPs; example of US tier-1 ISPs are: AT&T, GTT Communications and Lumen Technologies: a list on tier-1 ISPs.
  Tier-1 ISP are also known as Transit Providers, because they allow customer's internet traffic to cross or transit the transit provider network to connect to the rest of Internet. Transit Providers are also called upstream providers.
• regional (tier-2) ISPs are at the intermediate level.
  Since no tier-1 ISP is present in all the cities of the world, in some regions there exist a regional ISP which the access ISPs connect to; then each regional ISP is connected to a tier-1 ISP.
• access (tier-3) ISPs are at the bottom level.
  An access ISP purchases Internet transit from higher-tier ISPs and connects end users to Internet by charging them a fee. Tier-3 ISPs are considered last mile providers, because they do not have their own extensive network infrastructure. There are hundreds of thousands of access ISPs.

Each regional ISP is connected to a tier-1 ISP and pay the ISP which is connected to; each access ISP is connected to a regional ISP (but it may be connected to a tier-1 ISP) and pay the ISP which is connected to. Tier-1 ISPs are interconnected to each others, but do not pay anybody. The lower level ISPs are charged to connect to the higher level ISP and therefore are said to be a customer of the higher tier ISP, the higher tier ISP is said to be a provider.

A point of presence (PoP) is a group of multiple routers in the same place in the provider network, where customers can connect, PoP can be found at any levels of the hierarchy, except for the access ISP.

ISPs may multi-home, that is to connect to two or more providers; for example an access ISP may multi-home with two tier-2 ISPs and a tier-2 ISP may multi-home with multiple tier-1 ISPs. Multi-homing allows to trasmit packets even if one of its providers has a failure.

Customer ISPs pay their provider ISPs to get Internet interconnectivity. A pair of close ISPs that are at the same hierarchy level can peer to reduce the cost paid for the traffic they exchange with their providers; if two customer ISPs peer, they directly connect their networks, so that all the traffic between them passes through the direct connection instead of through intermediaries. Note that, when two ISPs peer, they do not pay each other; tier-1 ISPs peer with each others without paying. An Internet Exchange point (IXP) is a meeting place, built by a third-party company, where multiple ISPs can peer together. An IXP resides typically in a standalone building with its own switches.

A content provider network (CPN) is a network infrastructure that is responsible for hosting and delivering content to end users. It consists of data centers that store and distribute content. The main purpose of a CPN is to ensure that the content is available to end users.
Google is an example of these content-providers networks; Google has dozens of data centers located in the North and South America, Europe, Asia and Australia, where a small/big data center has a tens/hundreds of thousands of servers. In addition, Google has small data centers in located inside IXPs.
Google data centers are interconnected by the Google private TCP/IP network, which spreads all around the world but is separated from the public Internet. Note that the Google network only carries traffic from and to Google servers. Google private network tries to bypass the higher tiers of the Internet by peering with the lower tier network; nevertheless, the Google network also connects to the tier-1 ISPs, because many access ISPs can only be reached by transiting through tier-1 networks; and Goolge pay for the traffic exchanged with tier-1 IPSs. Content providers create their networks to better control on how their services are offered to end users.

Figure: Interconnection of ISPs

Summary

End systems are interconnected by multiple tiers of Internet Service Providers organized in a hierarchy. The tier 1 ISPs are at the top of the routing hierarchy and exchange traffic directly with each other. Tier 2 and Tier 3 ISPs buy Internet transit from tier-1 providers and they may also engage in peering. End users only access the Internet when they need to get information, they represent the bottom of the routing hierarchy.

4 Protocol Layers and Their Service Models

What is the Internet Network Architecture? How to design a network architectural model?

4.1 Layered Architecture

Modularity with Multi Layered Software Architecture

A layered software architecture allows to describe small and simple parts of a large and complex system. A layered architecture simplifies software development by bringing modularity, making easier to change the implementation of a service offered by one layer. A network developer can change the implementation of a layer, without having to change the other parts of the system, as long as the layer provides the layer above it with the same service and uses the same services from the layer below it.

Network Protocol Layering

Let’s consider network protocols. To structure the design of network protocols, network designers organized the protocols and their implementations in layers with each protocol belonging to one layer, like the functions of the architecture of the airline system.
Each layer offers to the layer above some services; this is called the service model of a layer. Just like the example of the airline, each layer provides its services in two ways: 1) by performing certain actions within that layer and 2) by using the services of the layer directly below it.

A protocol layer is implemented in software, in hardware or in a mix of software and hardware: the protocols of the application layer and of the transport layer are always implemented in software in the end systems; the network layer is often implemented by a mix of hardware and software; the physical layer and data link layer are commonly implemented in a network interface card, such as Ethernet or WiFi interface cards. Note that the functions in the layered architecture of the airline system were distributed among the many airports and flight control centers that constitute the system, so it is for a generic network protocol, which may be distributed among the end systems and packet switches that constitute the network.

Figure: The Internet protocol stack

The protocol Stack

The protocols of the layers, altogether, are called the protocol stack. The Internet protocol stack consists of five layers: physical, link or data-link, network, transport and application layers as illustrated in the figure above.

Application Layer

The Internet's application layer interacts directly with the network applications and provides them with the application-layer protocols that allow applications to communicate.

Application-layer protocols are distributed on multiple end systems; network applications use the application-layer protocols to exchange packets of information. The packets of information at the application layer are named messages. Example of Internet application layer protocols are: the HTTP protocol for transferring webpages, the SMTP protocol for transferring email messages, the FTP protocol for transferring files between two end-systems, the DNS protocol for translating human friendly domain names to network addresses.

Transport Layer

The Internet’s transport layer moves application-layer messages between application endpoints.

The TCP and UDP protocols are the two trasport-layer protocols that can transport application-layer messages and provide logical communication between application processes running on different hosts.
TCP protocol provides a connection-oriented service to its applications: it guarantees reliable in-sequence delivery of data, (segments are delivered in the same order in which they are sent); it provides flow control by matching the speed of sender and receiver; it provides for congestion control, so that a source reduces its transmission rate when network is congested.
UDP protocol provides a connectionless service to its application: it cannot offer reliability, flow control and congestion control.

Network Layer

The Internet's network-layer moves network-layer datagrams through a path of routers from the source host to the destination host, so providing a logical communication between hosts.
The TCP and UDP transport-layer protocols, in a source host, pass a transport-layer segment and a destination network address to the network layer, the network layer sends them to the network layer of the destination host.

The IP protocol is the most important Network Layer protocol. The IP protocol defines the fields in the datagram and how these fields are used by end systems and routers. All network-layer components must run the IP protocol.
The Routing protocols determine the route or path taken by datagrams as they flow from sources to destinations. There are many Internet routing protocols; the Internet is a network of networks and network administrators can run in their network the routing protocol of their choice.
Note: the network layer is often called as the IP layer (even though the network layer contains many other routing protocols than the the IP protocol) because the IP is the protocol that holds together computer networks.

Link Layer

As already said, the network layer moves a datagram through a series of routers. To accomplish this, the network layer uses the services of the link layer to move a packet from one node (host or router) to the next node in the route: in fact, at each node, the network layer passes a datagram down to the link layer, which delivers the datagram to the next node, where the link layer passes the datagram up to the network layer.

The link layer offers services which depends on the link-layer protocol used on the link. Some protocol provides reliable delivery by retransmitting over the link to the receiving node. Example of link-layer protocols are: Ethernet protocol, WiFi protocol, DOCSIS protocol.
Note: a network-layer datagram may need to traverse many links to go from a source host to destination host, therefore a datagram may be handled by different link-layer protocols in the different links through the route.
Note: link-layer packets are also called frames.

Physical Layer

Whereas the link layer is responsible for moving a frame from a transmitting node to a receiving node, the physical layer is responsible for moving the single bits within the frame from one node to the next.
The protocols in this layer depends on the link and on the type of the transmission medium. For example, Ethernet has several physical-layer protocols: one for twisted-pair cable, another for coaxial cable, another for fiber optic cable etc.

4.2 Encapsulation

Figure: each device contains a different set of layers

Figure above demonstrates that the data takes a physical path down a sending end system’s protocol stack, up and down the protocol stacks of link-layer switch and router, and then up the protocol stack at the destination end system.
Like end systems, packet switches (routers and link-layer switches) organize network protocols and implementations into layers, but packet switches do not implement all of the layers in the protocol stack: link-layer switches implement layers 1 and 2 and routers implement layers 1 to 3.
This implies that routers recognize layer 3 addresses such as IP address, whereas link-layer switches recognize layer 2 addresses such as Ethernet addresses, but cannot recognize IP addresses.
Note that all the complexity of the Internet structure is at its edge as the end systems implement all five layers of the protocol stack.

Encapsulation in the figure above

At the sending host, the application layer passes an application-layer message (M) to the transport layer.
The transport layer appends the transport-layer header information Ht. The header information include information that allows the transport layer, at the destination host, to deliver the message to the correct application, and error-detection bits. The application-layer message and the transport-layer header information together constitute the transport-layer segment.
The transport layer passes the segment to the network layer, which adds network-layer header information (Hn), for example network addresses of source and destination end systems, creating a network-layer datagram. The network layer passes the datagram to the link layer.
The link layer add the link-layer header information and create a link-layer frame.
At each layer, a packet has two different types of fields: header and a payload, where the payload is a packet from the above layer.

Note: the encapsulation process can be more complicated; for example, a big application message may be divided into multiple transport-layer segments, each of which may be divided into multiple network-layer datagrams.