Introduction to Packet-Switched Networks
Computer and communication networks provide a wide range of services, from simple networks of computers to remote-file access to digital libraries, voice over IP (VoIP), Internet gaming, cloud computing, video streaming and conferencing, television over Internet, wireless data communication, and networking billions of users and devices. Before exploring the world of computer and communication networks, we need to study the fundamentals of packet-switched networks as the first step. Packet-switched networks are the backbone of the data communication infrastructure. Therefore, our focus in this chapter is on the big picture and the conceptual aspects of this backbone highlighted as:
- Basic definitions in networks
- Types of packet-switched networks
- Packet size and optimizations
- Foundation of networking protocols
- Addressing scheme in the Internet
- Equal-sized packet model
We start with the basic definitions and fundamental concepts, such as messages, packets, and frames, and packet switching versus circuit switching. We learn what the Internet is and how Internet service providers (ISPs) are formed. We then proceed to types of packet-switched networks and how a message can be handled by either connection-oriented networks or connectionless networks. Because readers must get a good understanding of packets as data units, packet size and optimizations are also discussed.
We next briefly describe specific type of networks used in the Internet. Users and networks are connected together by certain rules called protocols. The Internet Protocol (IP), for example, is responsible for using prevailing rules to establish paths for packets. Protocols are represented by either the TCP/IP model or the OSI model. The five-layer TCP/IP model is a widely accepted Internet backbone protocol structure. In this chapter, we give an overview of these five layers and leave any further details to be discussed in the remaining chapters. Among these five layers, the basics of IP packets and network addressing are designated a separate section in this chapter, entitled IP Packets and Addressing. We make this arrangement because basic definitions related to this layer are required in the following few chapters.
As numerous protocols can be combined to enable the movement of packets, the explanation of all other protocols will be spread over almost all upcoming chapters. In the meantime, the reader is cautiously reminded that getting a good grasp of the fundamental material discussed in this chapter is essential for following the details or extensions described in the remainder of the book. At the end of this chapter, the equal-sized packet protocol model is briefly introduced.
1.1 Basic Definitions in Networks
Communication networks have become essential media for homes and businesses. The design of modern computer and communication networks must meet all the requirements for new communication applications. A ubiquitous broadband network is the goal of the networking industry. Communication services need to be available anywhere and anytime. The broadband network is required to support the exchange of multiple types of information, such as voice, video, and data, among multiple types of users, while satisfying the performance requirement of each individual application. Consequently, the expanding diversity of high-bandwidth communication applications calls for a unified, flexible, and efficient network. The design goal of modern communication networks is to meet all the networking demands and to integrate capabilities of networks in a broadband network.
Packet-switched networks are the building blocks of computer communication systems in which data units known as packets flow across networks. The goal of a broadband packet-switched network is to provide flexible communication in handling all kinds of connections for a wide range of applications, such as telephone calls, data transfer, teleconferencing, video broadcasting, and distributed data processing. One obvious example for the form of traffic is multi-rate connections, whereby traffic containing several different bit rates flows to a communication node. The form of information in packet-switched networks is always digital bits. This kind of communication infrastructure is a significant improvement over the traditional telephone networks known as circuit-switched networks.
1.1.1 Packet Switching Versus Circuit Switching
Circuit-switched networks, as the basis of conventional telephone systems, were the only existing personal communication infrastructures prior to the invention of packet-switched networks. In the new communication structure, voice and computer data are treated the same, and both are handled in a unified network known as a packet-switched network, or simply an integrated data network. In conventional telephone networks, a circuit between two users must be established for communication to occur. Circuit-switched networks require resources to be reserved for each pair of end users. This implies that no other users can use the already dedicated resources for the duration of network use and thus the reservation of network resources for each user may result in inefficient use of available bandwidth.
Packet-switched networks with a unified, integrated data network infrastructure collectively known as the Internet can provide a variety of communication services requiring different bandwidths. The advantage of having a unified, integrated data network is the flexibility to handle existing and future services with remarkably better performance and higher economical resource utilizations. An integrated data network can also derive the benefits of central network management, operation, and maintenance. Numerous requirements for integrated packet-switched networks are explored in later chapters:
- Having robust routing protocols capable of adapting to dynamic changes in network topology
- Maximizing the utilization of network resources for the integration of all types of services
- Providing quality of service to users by means of priority and scheduling
- Enforcing effective congestion-control mechanisms that can minimize dropping packets
Circuit-switched networking is preferred for real-time applications. However, the use of packet-switched networks, especially for the integration and transmission of voice and data, results in the far more efficient utilization of available bandwidth. Network resources can be shared among other eligible users. Packet-switched networks can span a large geographical area and comprise a web of switching nodes interconnected through transmission links. A network provides links among multiple users facilitating the transfer of information. To make efficient use of available resources, packet-switched networks dynamically allocate resources only when required.
1.1.2 Data, Packets, and Frames
A packet-switched network is organized as a multilevel hierarchy. In such a network, digital data are fragmented into one or more smaller units of data, each appended with a header to specify control information, such as the source and the destination addresses, while the remaining portion carries the actual data, called the payload. This new unit of formatted message is called a packet, as shown in Figure 1.1. Packets are forwarded to a data network to be delivered to their destinations. In some circumstances, packets may also be required to be attached together or further partitioned, forming a new packet having a new header. One example of such a packet is referred to as frame. Sometimes, a frame may be required to have more than one header to carry out additional tasks in multiple layers of a network.
Figure 1.1 Creating packets and frames out of a raw digital data
As shown in Figure 1.2, two packets, A and B, are being forwarded from one side of a network to the other side. Packet-switched networks can be viewed from either an external or an internal perspective. The external perspective focuses on the network services provided to the upper layers; the internal perspective focuses on the fundamentals of network topology, the structure of communication protocols, and addressing schemes.
Figure 1.2 A packet-switched network receiving various-sized packets to route out
A single packet may even be split into multiple smaller packets before transmission. This well-known technique is called packet fragmentation. Apart from measuring the delay and ensuring that a packet is correctly sent to its destination, we also focus on delivering and receiving packets in a correct sequence when the data is fragmented. The primary function of a network is directing the flow of data among the users.
1.1.3 The Internet and ISPs
The Internet is the collection of hardware and software components that make up our global communication network. The Internet is indeed a collaboration of interconnected communication vehicles that can network all connected communicating devices and equipment and provide services to all distributed applications. It is almost impossible to plot an exact representation of the Internet, since it is continuously being expanded or altered. One way of imagining the Internet is shown in Figure 1.3, which illustrates a big-picture view of the worldwide computer network.
Figure 1.3 The Internet, a global interconnected network
To connect to the Internet, users need the services of an Internet service provider (ISP). ISPs consist of various networking devices. One of the most essential networking devices is a router. Routers are network “nodes” that can operate to collectively form a network and to also connect ISPs together. Routers contain information about the network routes, and their tasks are to route packets to requested destinations.
Users, networking devices, and servers are connected together by communication links. Routers operate on the basis of one or more common routing protocols. In computer networks, the entities must agree on a protocol, a set of rules governing data communications and defining when and how two users can communicate with each other. Each country has three types of ISPs:
- National ISPs
- Regional ISPs
- Local ISPs
At the top of the Internet hierarchy, national ISPs connect nations or provinces together. The traffic between each two national ISPs is very heavy. Two ISPs are connected together through complex switching nodes called border routers (or gateway routers). Each border router has its own system administrator. In contrast, regional ISPs are smaller ISPs connected to a national ISP in a hierarchical chart. Each regional ISP can give services to part of a province or a city. The lowest networking entity of the Internet is a local ISP. A local ISP is connected to a regional ISP or directly to a national service provider and provides a direct service to end users called hosts. An organization that supplies services to its own employees can also be a local ISP.
Figure 1.4 illustrates a different perspective of the global interconnected network. Imagine the global network in a hierarchical structure. Each ISP of a certain hierarchy or tier manages a number of other network domains at its lower hierarchy. The structure of such networks resembles the hierarchy of nature from the universe to atoms and molecules. Here, Tier 1, Tier 2, and Tier 3 represent, respectively, a national ISP, a regional ISP, and a local ISP.
Figure 1.4 Hierarchy of networks from a different angle
1.1.4 Classification of ISPs
In most cases, a separate network managed by a network administrator is known as a domain, or an autonomous system. A domain is shown by a cloud in this book. Figure 1.5 shows several domains. An autonomous system can be administered by an Internet service provider (ISP). An ISP provides Internet access to its users. Networks under management of ISPs can be classified into two main categories: wide area networks (WANs) and local area networks (LANs). A wide area network can be as large as the entire infrastructure of the data network access system known as the Internet. A communication network can also be of wireless type both at LAN or WAN scales. We refer to such networks as wireless networks.
Figure 1.5 Overview of various types of Internet service providers (ISPs)
Figure 1.5 shows several major WANs each connected to several smaller networks such as a university campus network. Depending on the size of the network, a smaller network can be classified as a LAN or as a WAN. The major WANs are somehow connected together to provide the best and fastest communication for customers. One of the WANs is a wide area wireless network that connects wireless or mobile users to destination users. We notice that aggregated traffic coming from wireless equipment such as smartphone and a mobile laptop in the wide area wireless network is forwarded to a link directed from a major node. The other WAN is the telephone network known as public-switched telephone network (PSTN) that provides telephone services.
As an example of the local area network, a university campus network is connected to the Internet via a router that connects the campus to an Internet service provider. ISP users from a residential area are also connected to an access point router of the wide area ISP, as seen in the figure. Service providers have varying policies to overcome the problem of bandwidth allocations on routers. An ISP’s routing server is conversant with the policies of all other service providers. Therefore, the “ISP server” can direct the received routing information to an appropriate part of the ISP. Finally, on the left side of Figure 1.5, we see the data center network connected to the wide area packet-switched network. Cloud computing data centers contain databases and racks of servers that provide brilliant data processing services; these are discussed in detail in Chapter 16.
Network nodes (devices) such as routers are key components that allow the flow of information to be switched over other links. When a link failure occurs in a packet-switched network, the neighboring routers share the fault information with other nodes, resulting in updating of the routing tables. Thus, packets may get routed through alternative paths bypassing the fault. Building the routing table in a router is one of the principal challenges of packet-switched networks. Designing the routing table for large networks requires maintaining data pertaining to traffic patterns and network topology information.