Networking, Internet and Distributed Computing

Introduction

Modern information systems rarely operate as isolated computers. They depend on networks that transport data, identify communicating devices, provide shared services, and coordinate computation across locations. A banking transaction, video conference, cloud application, industrial sensor network, and online learning platform all rely on related principles: layered communication, addressing, routing, reliable transport, security, service discovery, and distributed coordination.

This chapter develops those principles from the network model to practical diagnosis and system design. It also examines wireless networking, Internet services, distributed architectures, and the professional responsibilities associated with operating connected systems.

Learning Objectives

By the end of this chapter, you should be able to:

1. Explain the purpose and operation of the OSI and TCP/IP network models.
2. Describe Ethernet, IP, TCP, UDP, DNS, HTTP, DHCP, and related protocols.
3. Calculate IPv4 subnet ranges and distinguish public, private, static, and dynamic addressing.
4. Explain routing, switching, wireless technologies, and network performance factors.
5. Compare client–server, peer-to-peer, cloud, edge, and microservices architectures.
6. Apply systematic techniques to diagnose network faults.
7. Design a secure, scalable, and maintainable network for an organisational scenario.
8. Evaluate reliability, security, governance, and ethical issues in distributed systems.

1. Networks as Communication Systems

A computer network is a collection of interconnected devices that exchange data using agreed rules called protocols. Devices may include clients, servers, switches, routers, access points, firewalls, sensors, and virtual machines. A network provides several fundamental services:

• Communication: transfer of messages between endpoints.
• Resource sharing: access to files, printers, applications, databases, and Internet connections.
• Availability: continued service through redundancy and failover.
• Scalability: addition of users, devices, and applications without unacceptable degradation.
• Management: monitoring, authentication, configuration, and policy enforcement.

Network performance is commonly discussed using different measures. Bandwidth is the maximum data-carrying capacity of a link, measured in bits per second. Throughput is the actual rate achieved. Latency is the time required for data to travel between points. Jitter is variation in delay, which is particularly important for voice and video. Packet loss occurs when transmitted packets do not reach their destination.

A useful idealised relationship is:

The inequality is not an equation for predicting exact performance. It states that actual throughput cannot exceed the capacity of the path and is reduced by protocol overhead, congestion, interference, processing limitations, and loss. For a file of size S bits transferred at throughput R bits per second, the ideal transmission time is:

This assumes that R remains constant and ignores propagation delay, acknowledgements, headers, retransmissions, and application processing.

2. Layered Network Models

Layering divides a complex communication problem into manageable functions. Each layer provides services to the layer above and uses services from the layer below. Encapsulation occurs when each layer adds control information to data as it moves down the sender’s protocol stack. At the receiver, decapsulation removes those headers.

The OSI model

The seven-layer Open Systems Interconnection model is a conceptual reference framework:

The OSI model is valuable for learning and troubleshooting, although actual Internet protocols do not always map neatly to one layer. For example, TLS operates between application protocols and transport services and may be described using different layer interpretations.

The TCP/IP model

The practical Internet architecture is often represented by four layers:

1. Link: local network technologies such as Ethernet and Wi-Fi.
2. Internet: IP addressing and routing.
3. Transport: process-to-process delivery using TCP or UDP.
4. Application: services such as web, naming, email, and file transfer.

The models are not competing implementations. OSI is mainly a pedagogical and analytical model; TCP/IP describes the protocol family used by the Internet.

3. Local Networks, Switching and Media Access

Ethernet sends data in frames across local networks. A frame contains source and destination Media Access Control (MAC) addresses, a payload, and error-detection information. A switch learns which MAC address is reachable through each port and forwards frames selectively. Unknown unicast, broadcast, and some multicast traffic may be flooded within the relevant broadcast domain.

A collision domain is a network area in which simultaneous transmissions can interfere. Modern switched full-duplex Ethernet largely eliminates collisions on ordinary switch links. A broadcast domain is the set of devices that receive a Layer 2 broadcast. Virtual local area networks (VLANs) divide a physical switching infrastructure into separate logical broadcast domains. Inter-VLAN communication requires a router or Layer 3 switch.

Common media include:

• Twisted-pair copper: economical and widely used for office access networks; susceptible to distance and electromagnetic limitations.
• Fibre optic: high capacity, long distance, and resistance to electromagnetic interference; generally requires more specialised installation.
• Radio: enables mobility but is affected by distance, obstacles, interference, spectrum contention, and security configuration.

4. IP Addressing and Subnetting

An IP address identifies an interface within an IP network. IPv4 uses 32 bits, normally written as four decimal octets, such as 192.168.10.25. IPv6 uses 128 bits and is written in hexadecimal groups, such as 2001:db8:1::25. IPv6 provides a vastly larger address space and supports autoconfiguration and modern network design, although IPv4 remains extensively deployed.

An IPv4 address consists of a network prefix and a host portion. CIDR notation writes the prefix length after a slash. For example, /24 means that 24 of the 32 bits identify the network.

For a conventional IPv4 subnet with prefix length p, the number of addresses is:

where N is the total number of addresses and p is between 0 and 32. In the traditional case, two addresses are reserved: the network address and the directed broadcast address. Therefore, usable host addresses are:

This host formula is not universally applicable: /31 point-to-point networks and /32 host routes are important exceptions, and IPv6 does not use an IPv4-style broadcast address.

Worked example: subnet interpretation

Consider 192.168.20.0/26. The prefix length is 26, so the host portion contains 6 bits.

Under conventional IPv4 subnet assumptions:

The block size is 64 addresses. The range is:

• Network address: 192.168.20.0
• Usable hosts: 192.168.20.1 through 192.168.20.62
• Broadcast address: 192.168.20.63

The next /26 block begins at 192.168.20.64. Subnetting reduces broadcast scope and allows address allocation to reflect organisational structure. However, excessively small subnets can waste addresses and complicate administration.

Private IPv4 ranges defined for internal use include 10.0.0.0/8, 172.16.0.0/12, and 192.168.0.0/16. Private addresses are not globally routable and commonly reach the Internet through Network Address Translation (NAT). NAT conserves public IPv4 addresses but can complicate end-to-end connectivity, logging, and some application protocols.

5. Routing and Internet Protocols

A router forwards packets between IP networks. It examines the destination address and selects a route using a routing table. Routes may be directly connected, statically configured, or learned through dynamic routing protocols.

Routers use longest-prefix matching: when multiple routes match a destination, the route with the greatest prefix length is normally preferred because it is the most specific. A default route, commonly written as 0.0.0.0/0 for IPv4, is used when no more specific route exists.

Routing protocols have different purposes. RIP uses a simple distance-vector approach and is limited in larger environments. OSPF is a link-state interior gateway protocol that builds a topology database and calculates paths using a shortest-path method. BGP exchanges reachability information between autonomous systems and is fundamental to Internet-scale interconnection. Route selection also considers administrative policy, not merely physical distance.

The Internet Protocol provides best-effort packet delivery. IP does not guarantee delivery, ordering, or absence of duplication. These responsibilities, when required, are handled by higher layers.

TCP and UDP

TCP provides a connection-oriented byte stream. It uses sequence numbers, acknowledgements, retransmission, flow control, and congestion control. A TCP connection is established using a three-way handshake: SYN, SYN-ACK, and ACK. TCP is appropriate for applications such as web transactions, secure shell, and file transfer where reliable ordered delivery is important.

UDP is connectionless and provides datagrams with minimal overhead. It does not itself guarantee delivery, ordering, or congestion control. It is useful for DNS queries, real-time media, online games, and protocols that implement their own reliability or prioritise low delay.

A port number identifies an application endpoint on a host. An IP address identifies the host interface; the combination of address and port helps identify a network service. A socket is commonly described as an IP address and port associated with a transport protocol.

Essential Internet services

• DNS: translates names such as example.org into resource records, including IP addresses. DNS caching improves performance but means changes may not be visible everywhere immediately.
• DHCP: automatically supplies configuration such as an IP address, subnet mask or prefix, default gateway, and DNS server.
• HTTP and HTTPS: support web communication. HTTPS combines HTTP with TLS to provide confidentiality, integrity, and server authentication when correctly configured.
• Email protocols: SMTP transfers mail between servers and from clients to submission services; IMAP and POP3 support mailbox access with different synchronisation models.
• NTP: helps synchronise system clocks. Accurate time is important for logs, authentication, certificates, and distributed coordination.
• ICMP: communicates network control and diagnostic information. Tools such as ping and traceroute use ICMP or related mechanisms, although firewalls may filter diagnostic packets.

6. Wireless Networking

Wi-Fi uses radio communication and a shared medium. Unlike a dedicated full-duplex cable, wireless transmission is affected by contention, signal strength, channel overlap, attenuation, and hidden nodes. Wi-Fi networks use an access point and client stations, although ad hoc and mesh arrangements also exist.

Important design considerations include:

• selecting channels with limited interference;
• providing overlapping coverage without excessive co-channel contention;
• placing access points according to measured signal and capacity requirements;
• separating guest, staff, operational, and management traffic using VLANs or equivalent controls;
• using current security standards such as WPA3 where supported;
• disabling insecure legacy authentication and changing default administrative credentials;
• planning for roaming, device density, voice traffic, and power availability.

Wireless performance should not be estimated from signal strength alone. A strong signal can coexist with congestion or interference. A site survey, spectrum analysis, access-point statistics, and application-level tests provide better evidence.

7. Network Security and Governance

Network security protects confidentiality, integrity, availability, authenticity, and accountability. A layered approach is more reliable than depending on one firewall.

Key controls include:

• firewalls and security groups that enforce least-privilege traffic rules;
• network segmentation separating users, servers, guests, and sensitive devices;
• strong identity management, multifactor authentication, and role-based access;
• encryption in transit and, where appropriate, at rest;
• secure configuration baselines, patch management, and vulnerability assessment;
• intrusion detection and prevention;
• centralised logging, time synchronisation, and alerting;
• tested backups and incident-response procedures.

A firewall should be designed from explicit policy rather than a collection of unexplained exceptions. Rules should identify source, destination, service, direction, justification, owner, and review date. Zero-trust principles require continuous verification and do not assume that an internal network is automatically trustworthy.

Network monitoring raises governance and privacy questions. Organisations should define legitimate purposes, retention periods, access controls, and disclosure procedures for packet metadata, user activity, and security logs. Monitoring employees or students without transparent policy may violate law, institutional rules, or professional ethics. Data protection regulations may impose requirements for minimisation, lawful processing, breach reporting, and cross-border data transfer.

8. Distributed Computing Architectures

Distributed computing divides processing, storage, or coordination across multiple networked machines. The machines cooperate by exchanging messages, yet each has independent failure modes and may have incomplete knowledge of the overall system.

Architectural styles

Client–server: clients request services from central or clustered servers. It is easy to manage but may create bottlenecks or central points of failure.

Peer-to-peer: nodes may act as both clients and servers. This can improve resilience and resource utilisation but makes trust, discovery, and management more difficult.

Three-tier architecture: presentation, application logic, and data management are separated. This supports maintainability and independent scaling.

Microservices: an application is decomposed into relatively independent services communicating through APIs or messaging. Benefits include team autonomy and selective scaling; costs include operational complexity, network failure, distributed tracing, data consistency challenges, and deployment coordination.

Cloud computing: providers offer infrastructure, platforms, or software as services. Elasticity and rapid provisioning are advantages, but organisations must manage identity, configuration, cost, data location, vendor dependence, and shared-responsibility boundaries.

Edge computing: processing occurs close to data sources or users, reducing latency and bandwidth consumption. Edge nodes introduce additional physical, security, and lifecycle-management challenges.

Distributed-system concerns

Networks introduce delay, loss, duplication, partition, and reordering. A remote call is not equivalent to a local function call. Applications should define timeouts, retries, idempotency, circuit breakers, and graceful degradation. Unbounded retries can amplify an outage; exponential backoff with jitter is commonly used to reduce synchronised retry storms.

Distributed systems often discuss consistency, availability, and partition tolerance. The CAP principle states that when a network partition occurs, a system cannot simultaneously provide both perfect consistency and full availability for all operations. This is a statement about guarantees during partitions, not a claim that systems permanently choose only two qualities. Designers must specify which operations may return stale data, reject requests, or use conflict resolution.

Replication improves availability and read performance but creates consistency and conflict-management requirements. Consensus protocols such as Raft help a group of nodes agree on an ordered log under defined failure assumptions. They do not eliminate network delay or make arbitrary failures harmless.

9. Network Design Methodology

A professional design begins with requirements rather than equipment selection.

1. Gather requirements: users, applications, traffic patterns, security obligations, growth, availability targets, budget, and support skills.
2. Document constraints: building structure, cabling routes, spectrum, legacy systems, regulatory requirements, and cloud dependencies.
3. Create a logical design: addressing plan, VLANs, routing domains, trust zones, naming, authentication, and service placement.
4. Create a physical design: switches, routers, firewalls, access points, links, racks, power, cooling, and redundancy.
5. Model capacity: estimate peak traffic, concurrent users, wireless density, storage traffic, and growth. Avoid designing only for average demand.
6. Build security in: least privilege, segmentation, secure management, logging, backups, and recovery testing.
7. Validate: use a pilot, acceptance tests, failure simulations, and performance measurements.
8. Operate: maintain diagrams, configurations, asset records, change control, monitoring, and review cycles.

High availability may use redundant links, devices, power supplies, or sites. Redundancy is useful only when failover is tested and operational dependencies are understood. A redundant firewall with a single unmonitored power circuit is not a fully resilient design.

10. Diagnostic Techniques

Troubleshooting should proceed from evidence, scope, and simple tests toward deeper analysis. A useful sequence is:

1. Define the symptom, affected users, time, and recent changes.
2. Determine whether the fault is isolated, local, subnet-wide, service-wide, or Internet-wide.
3. Check physical links, power, interface state, and wireless association.
4. Inspect IP configuration, address assignment, gateway, DNS, and routes.
5. Test progressively: local host, local gateway, remote IP address, DNS resolution, and application port.
6. Compare a working device with a failing device.
7. Inspect switch, router, firewall, server, and authentication logs.
8. Capture packets when protocol behaviour is uncertain.
9. Apply one controlled change at a time and document the result.
10. Confirm restoration and identify the root cause, not merely the symptom.

Useful tools include ipconfig or ip, ping, traceroute or tracert, nslookup or dig, arp, route, ss or netstat, and Wireshark. Tool output must be interpreted carefully. A failed ping does not prove that a service is unavailable because ICMP may be blocked; a successful ping does not prove that an application port or authentication service works.

Common fault patterns include duplicate IP addresses, incorrect VLAN assignment, exhausted DHCP pools, invalid DNS records, asymmetric routing, MTU mismatch, certificate expiration, firewall rule errors, wireless interference, and overloaded links. Good documentation reduces mean time to repair and prevents recurrence.

11. Applied Case: Illustrative Campus Network Upgrade

Background

A fictional technical college has 1,200 students, 140 staff members, laboratories, administrative offices, a learning-management system, voice-over-IP phones, and guest Wi-Fi. Users report slow wireless access, intermittent access to the learning system, and difficulty reaching printers across buildings.

Problem analysis

The existing network uses one large flat subnet. Broadcast traffic is unnecessary for many devices, guest users share the staff network, access points use overlapping channels, and the DHCP pool is nearly exhausted. The Internet gateway is a single device with limited logging. Troubleshooting is difficult because switch, firewall, and authentication logs are not centrally collected.

Approach and solution

The design team first maps applications and traffic, performs a wireless survey, and inventories devices. It creates separate VLANs for administration, staff, students, laboratories, voice, servers, management, and guests. Inter-VLAN routing is placed on redundant Layer 3 switches, while a firewall controls traffic between trusted zones and the Internet. Guest traffic is Internet-only and rate-limited.

The addressing plan uses appropriately sized private IPv4 subnets and reserves space for growth. DHCP is centralised with redundant services; DNS records are reviewed; network devices use secure management protocols and multifactor authentication where supported. Access points are repositioned and channels are planned using survey evidence. WPA3 or an appropriately secured enterprise wireless configuration is deployed.

Monitoring collects interface utilisation, authentication events, DHCP failures, DNS health, firewall decisions, and wireless client statistics. The team tests loss of a switch link, failure of a DHCP server, access-point overload, and firewall failover before production migration.

Outcome

The segmented design reduces broadcast scope, prevents guests from reaching internal systems, and makes policy easier to audit. Wireless capacity improves because channel use and access-point placement are based on measured demand. Central logs shorten diagnosis time. The case also demonstrates that performance problems are often architectural and operational, not simply caused by insufficient Internet bandwidth.

Learning points

A successful network project connects requirements, security, addressing, wireless engineering, monitoring, and change management. Technical improvements must be accompanied by documentation, staff training, user communication, and tested recovery procedures.

12. Contemporary Developments and Best Practice

IPv6 adoption, software-defined networking, network function virtualisation, 5G, low-power IoT networks, service meshes, and programmable observability are reshaping network operations. Artificial intelligence and machine learning can assist anomaly detection, capacity forecasting, and log correlation, but models may generate false positives, inherit biased training data, or conceal reasoning. Automated remediation should therefore use limited privileges, approval thresholds, rollback mechanisms, and audit trails.

Best practice is evidence-driven and lifecycle-oriented:

• use open standards and documented interfaces where practical;
• maintain accurate topology, addressing, asset, and dependency records;
• automate repeatable configuration with review and version control;
• measure user experience, not only device health;
• design for failure and test recovery;
• minimise privileges and exposed services;
• patch according to risk and verify configuration drift;
• protect secrets and certificates throughout their lifecycle;
• review suppliers, cloud contracts, data location, and exit strategies;
• train operators and communicate planned changes.

Chapter Summary

Networks provide structured communication among devices and services. Layered models explain how applications, transport protocols, IP, link technologies, and physical media cooperate through encapsulation. Ethernet and Wi-Fi support local connectivity; switches forward frames, while routers forward packets between IP networks. CIDR addressing and subnetting organise address space and reduce broadcast scope. TCP supplies reliable ordered delivery, whereas UDP provides lightweight datagrams. DNS, DHCP, HTTP(S), email protocols, NTP, and ICMP support essential Internet operations.

Effective network design considers requirements, capacity, segmentation, security, wireless conditions, redundancy, monitoring, and lifecycle management. Distributed computing extends these principles across multiple machines, where latency, failure, consistency, replication, and coordination must be explicitly managed. Systematic diagnosis uses scope, layered tests, logs, packet captures, and controlled changes. Professional practice also requires privacy protection, responsible monitoring, governance, documentation, and tested recovery.

Key Terms

Network: A set of interconnected devices that exchange data using protocols.

Protocol: An agreed rule set for communication.

Encapsulation: Adding protocol control information as data moves down a protocol stack.

MAC address: A link-layer identifier associated with a network interface.

Switch: A device that forwards frames within a local network.

Router: A device that forwards packets between IP networks.

VLAN: A logical Layer 2 broadcast domain implemented over switching infrastructure.

IP address: A logical address used for delivery across IP networks.

CIDR: Classless Inter-Domain Routing notation that represents an address and prefix length.

Subnet mask: A representation of which IPv4 bits identify the network and which identify hosts.

NAT: Translation between address spaces, commonly private and public IPv4 addresses.

TCP: A connection-oriented transport protocol providing reliable ordered byte delivery.

UDP: A connectionless transport protocol providing datagrams with minimal overhead.

DNS: The distributed naming system that maps domain names to resource records.

DHCP: A protocol for automatic host configuration.

Latency: The time taken for data to travel between communicating points.

Distributed system: A system whose components execute on multiple networked machines.

Replication: Maintaining copies of data or services on multiple nodes.

Zero trust: A security approach that continuously verifies access rather than trusting network location.

Observability: The ability to understand system state through metrics, logs, traces, and related evidence.

Review Questions

Short Answer Questions

1. What is the primary purpose of layering in network architecture?
2. Distinguish bandwidth, throughput, latency, and jitter.
3. What is the function of a default route?
4. Give two differences between TCP and UDP.
5. Why is network segmentation important for security and administration?

Long Answer Questions

1. Compare the OSI and TCP/IP models and explain why neither should be treated as a complete physical description of every modern protocol.
2. Explain IPv4 subnetting and calculate the network, usable host range, and broadcast address for 192.168.50.128/27.
3. Describe how DNS, DHCP, routing, and HTTPS cooperate when a user opens a secure website by name.
4. Compare client–server, peer-to-peer, cloud, edge, and microservices architectures, including their principal trade-offs.
5. Develop a network security and governance plan for an organisation containing staff, guests, servers, wireless users, and remote workers.

Analytical / Application Questions

1. A campus reports that users can ping a server but cannot access its web application. Construct a layered diagnostic plan and identify at least four possible causes.
2. Design a logical network for a small hospital with clinical devices, administration, visitors, voice, and medical servers. Justify segmentation, addressing, routing, monitoring, and access controls.
3. A distributed application experiences repeated timeouts during a network partition. Analyse how retries, consistency requirements, circuit breakers, replication, and graceful degradation should be designed.

Multiple Choice Questions with Answers

Question 1

Which device primarily forwards packets between different IP networks?

A. Hub

B. Router

C. Wireless client

D. Patch panel

Correct Answer: B. Router

Question 2

How many total IPv4 addresses are in a conventional /26 subnet?

A. 32

B. 64

C. 128

D. 256

Correct Answer: B. 64

Question 3

Which protocol translates domain names into resource records such as IP addresses?

A. DHCP

B. DNS

C. NTP

D. ICMP

Correct Answer: B. DNS

Question 4

Which transport protocol provides reliable, ordered byte delivery?

A. UDP

B. IP

C. TCP

D. ARP

Correct Answer: C. TCP

Question 5

A VLAN primarily creates what kind of separation?

A. Separate logical broadcast domains

B. Separate electrical power circuits

C. Separate CPU instruction sets

D. Separate file formats

Correct Answer: A. Separate logical broadcast domains

Question 6

Which statement about UDP is correct?

A. It guarantees ordered delivery

B. It requires a three-way handshake

C. It provides lightweight datagrams without built-in delivery guarantees

D. It replaces IP routing

Correct Answer: C. It provides lightweight datagrams without built-in delivery guarantees

Question 7

What is the main purpose of DHCP?

A. Encrypting web traffic

B. Automatically providing network configuration

C. Detecting malware in email

D. Compressing packet payloads

Correct Answer: B. Automatically providing network configuration

Question 8

Why can a successful ping fail to prove that a web application is available?

A. Ping tests only the relevant web application database

B. ICMP reachability does not prove that the application port and service work

C. Ping always uses TCP port 443

D. Ping can only operate on wireless networks

Correct Answer: B. ICMP reachability does not prove that the application port and service work

Question 9

Which architecture places processing close to data sources to reduce latency and bandwidth use?

A. Edge computing

B. Batch printing

C. Cold standby only

D. Single-tier computing

Correct Answer: A. Edge computing

Question 10

In a distributed system, what is a key danger of unlimited automatic retries?

A. They always improve availability without cost

B. They can amplify an outage through retry storms

C. They prevent all network partitions

D. They eliminate the need for timeouts

Correct Answer: B. They can amplify an outage through retry storms