Comprehensive undergraduate notes, network architecture diagrams, applied case studies and university assessment.
Undergraduate level · Chapter 7 of 16 · Network infrastructure and cloud platforms
Learning outcomes: Explain network models and protocols; plan addressing and routing; evaluate wireless and wired networks; apply network security and troubleshooting; compare cloud service and deployment models; design secure, resilient and cost-aware cloud solutions.
Communication networks and information systems
A computer network is a collection of interconnected devices that exchange data and share services through communication links and agreed protocols. In an information system, networks connect users, application servers, databases, sensors, branches, cloud platforms and external partners. Network quality influences whether business processes can operate reliably, securely and at acceptable cost. A network should therefore be designed around application requirements and organisational risk rather than bandwidth alone.
A network can be described by its nodes, links, interfaces, protocols, topology and services. Nodes include end-user devices, switches, routers, firewalls, wireless access points, servers and network appliances. Links may use copper, fibre, radio or satellite. Protocols define message formats, addressing, sequencing, error handling and other communication rules. The design must account for capacity, latency, jitter, packet loss, availability and security.
Networks are often classified by geographic scope. A personal area network connects devices around an individual; a local area network serves a home, office or campus; a metropolitan area network spans a town or city; and a wide area network connects distant locations. These categories describe scale, not necessarily ownership. A modern organisation may combine private networks, carrier services, public Internet access and cloud connectivity.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Network topologies and physical media
Topology describes how network components are arranged physically or logically. A star topology connects devices to a central switch or access point; a bus uses a shared communication medium; a ring connects nodes in a loop; and a mesh provides multiple paths between nodes. Ethernet LANs commonly use switched star-like physical layouts, while wireless networks share radio resources through access points.
A star is easy to manage and isolate, but the central device can become a failure point unless it is redundant. Mesh topologies can improve resilience by providing alternate paths, but increase cabling, configuration and cost. A bus is simple but shared-medium contention and fault isolation can be difficult. Logical topology may differ from the physical layout because VLANs, tunnels and routing create virtual relationships.
Copper twisted-pair cabling is widely used for Ethernet access. Fibre optic cable supports high bandwidth and longer distances with immunity to electromagnetic interference. Wireless radio enables mobility but is affected by distance, obstacles, interference and shared spectrum. Selection should consider distance, throughput, environmental conditions, installation cost, maintenance, security and future expansion.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
The OSI and TCP/IP reference models
The OSI reference model divides network communication into seven conceptual layers: physical, data link, network, transport, session, presentation and application. It provides a vocabulary for discussing responsibilities and troubleshooting. The TCP/IP model groups functions into link, Internet, transport and application layers and more closely reflects the protocol suite used on the Internet.
At the physical layer, signals and media carry bits. The data-link layer handles local frame delivery and media access. The network layer provides logical addressing and routing. The transport layer supports end-to-end communication, including reliability or datagram delivery. Application protocols provide services such as web access, name resolution, email and file transfer. Encapsulation adds protocol information as data moves down a stack.
Layering simplifies design because a protocol at one layer can use services below it without implementing every lower-level detail. However, real systems sometimes cross layers for performance, security or troubleshooting. A DNS failure, for example, may look like an application outage even when IP connectivity is healthy. Layer models guide diagnosis, but administrators must trace the complete path.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Ethernet, switching and local area networks
Ethernet is a widely deployed family of wired LAN technologies. Devices communicate using frames and hardware addresses on a local network segment. Modern Ethernet switches learn source MAC addresses and build forwarding tables, allowing frames to be sent to the appropriate port rather than repeating every frame to every connected device.
A switch reduces collisions compared with legacy shared Ethernet hubs and supports full-duplex communication. Broadcast traffic remains important because some protocols use broadcasts for local discovery. Excessive broadcast domains can affect performance and security, so network segmentation is commonly used. Switch configuration includes port speed, duplex, VLAN membership, trunking, spanning-tree protection and access controls.
A campus network often uses access switches for end devices, distribution or aggregation layers for policy and routing, and a core layer for high-capacity connectivity. Smaller environments may combine roles. Redundancy should avoid loops, and monitoring should detect port errors, congestion, unusual broadcasts and device failures. Physical security of wiring closets and switch ports is also essential.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
IP addressing and subnetting
Internet Protocol provides logical addressing so packets can be routed between networks. IPv4 uses 32-bit addresses, commonly written in dotted decimal notation; IPv6 uses 128-bit addresses and a hexadecimal representation. Addresses identify interfaces within an addressing plan. A subnet prefix indicates which bits represent the network portion and which identify addresses within that subnet.
Subnetting divides a larger address block into smaller networks. It supports efficient allocation, broadcast-domain control and policy boundaries. For example, a /24 IPv4 prefix contains 256 total addresses, although conventional network and broadcast addresses are not assigned to hosts on a standard subnet. A /26 contains 64 total addresses, typically 62 usable host addresses in conventional IPv4 subnetting.
A good addressing plan reserves ranges for infrastructure, users, servers, voice, guest wireless, management and point-to-point links. Avoid overlapping private address ranges across connected sites because overlap complicates routing and VPN connectivity. IPv6 planning should use hierarchical prefixes and support automatic configuration where appropriate. Documentation must track assignments and ownership.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Routing and internetworking
Routing moves packets between different IP networks. Routers examine destination addresses and consult routing tables to select a next hop or outgoing interface. A default route forwards traffic when no more specific route matches. Static routes are configured manually; dynamic routing protocols exchange reachability information and adapt to topology changes.
Routing decisions depend on prefix matching and route preference. Longest-prefix matching selects the most specific matching route. Dynamic protocols such as OSPF operate within an autonomous system, while BGP exchanges routing information between autonomous systems. Protocol choice depends on network scale, policy, convergence needs and operational expertise.
Routing failures can arise from incorrect masks, missing routes, asymmetric paths, filtering, address conflicts or unstable links. Troubleshooting should verify local interface status, address configuration, gateway reachability, route tables and path behaviour. Redundant routes need careful policy to avoid loops and unexpected traffic paths.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Transport protocols: TCP and UDP
The transport layer provides communication between application processes. Transmission Control Protocol (TCP) establishes a connection and provides ordered, reliable byte-stream delivery using acknowledgements, sequence numbers, retransmission and flow control. Applications use TCP when reliable ordered delivery is important, including many web, email and file transfer interactions.
User Datagram Protocol (UDP) sends datagrams without establishing a connection or guaranteeing delivery, ordering or retransmission. Its lower overhead can suit real-time voice, video, DNS and applications that implement their own recovery. UDP is not inherently faster in every context; performance depends on network conditions, congestion control and application design.
TCP uses a handshake to establish connection state and mechanisms to detect loss and manage flow. Congestion control reduces the risk that senders overwhelm the network. UDP applications must decide how to handle loss, duplication and timing. Administrators should understand ports, firewall policy and transport behaviour when diagnosing connection failures.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Wireless networking and Wi-Fi
Wireless LANs use radio communication to connect devices without a physical cable to each endpoint. Wi-Fi is based on IEEE 802.11 standards, with generations differing in frequency bands, channel widths, modulation and capabilities. Actual throughput is affected by signal strength, interference, contention, client capability, access-point placement and backhaul capacity.
Wireless planning includes a site survey, coverage objectives, channel selection, transmit power, roaming behaviour and capacity per access point. More access points do not always improve service: overlapping channels and excessive power can increase interference. Dense classrooms, lecture halls and offices need capacity planning based on concurrent clients and application usage.
Security should use current enterprise-grade authentication and encryption options supported by the environment. Guest access should be isolated from internal systems. Disable obsolete protocols, change default credentials, protect management interfaces and monitor rogue access points. Wireless networks should be treated as an extension of the organisational security perimeter, not as inherently trusted.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Network services: DNS, DHCP and NTP
Domain Name System translates names into resource records, allowing users and applications to locate services without memorising addresses. DNS records include A and AAAA for address mapping, CNAME for aliases, MX for mail exchange, TXT for text-based verification and service records for service discovery. Recursive resolvers cache answers according to TTL values.
Dynamic Host Configuration Protocol automatically supplies clients with settings such as IP address, subnet mask, default gateway and DNS resolver. DHCP scopes must avoid conflicts with static assignments and reserve addresses for infrastructure. Lease duration influences address reuse and client renewal behaviour. Rogue DHCP servers can redirect users or disrupt connectivity, so network access controls matter.
Network Time Protocol synchronises clocks across systems. Accurate time supports event correlation, authentication, certificates, logging and transaction analysis. Organisations should use trusted time sources and monitor drift. DNS, DHCP and NTP are foundational dependencies: failures may appear as application problems even when servers and links are functioning.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Network security architecture and segmentation
Network security protects the confidentiality, integrity and availability of communications and connected systems. Controls include firewalls, network access control, segmentation, secure remote access, intrusion detection, intrusion prevention, secure configuration, monitoring and physical protection. No single control is sufficient because threats can originate inside and outside an organisation.
Segmentation separates systems according to function, sensitivity and trust. VLANs can create separate Layer 2 broadcast domains, while routing and firewall rules control traffic between segments. A guest wireless network should not have unrestricted access to finance or student records. Administrative interfaces should be reachable only from approved management networks and authorised devices.
A zero-trust approach avoids granting implicit trust solely because a device is inside a network perimeter. Access decisions consider identity, device posture, resource sensitivity and context. Least privilege and continuous verification reduce lateral movement. Segmentation must be tested: a VLAN by itself does not guarantee isolation if routing, trunking or firewall rules are misconfigured.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Firewalls, VPNs and secure remote access
A firewall enforces traffic policy between network zones. Packet-filtering firewalls evaluate addresses, ports and protocols; stateful firewalls track connection state; application-aware firewalls inspect higher-level traffic; and web application firewalls focus on HTTP-based threats. Rules should be explicit, minimal, documented and reviewed regularly.
A virtual private network creates an encrypted tunnel across an untrusted network. Site-to-site VPNs connect networks; remote-access VPNs connect individual users or devices. VPN security depends on strong authentication, modern cryptography, endpoint security, routing policy and timely patching. A VPN does not make a compromised endpoint safe or justify broad access to every internal system.
Remote access should use multi-factor authentication, managed devices where possible, least-privilege access and logging. Privileged access may require a separate jump host or privileged access management process. Revoke access promptly when a role changes or employment ends. Test emergency access and recovery procedures without weakening everyday controls.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Network monitoring and troubleshooting
Network operations require visibility into availability, utilisation, latency, packet loss, interface errors, routing changes, wireless quality and security events. Monitoring may use SNMP, flow records, syslog, packet capture, synthetic probes and platform telemetry. Dashboards should connect technical measures to service outcomes, such as whether students can access a learning platform.
Troubleshooting follows a structured process: define the symptom, establish scope, identify recent changes, test hypotheses, isolate the layer or segment, implement a controlled fix and verify recovery. Tools include ping, traceroute, nslookup or dig, ipconfig or ip, route inspection, port tests and packet analysers. A successful ping does not prove that an application or its dependencies work.
Maintain network diagrams, IP plans, device inventories, configuration backups and incident records. Baselines help distinguish normal variation from an anomaly. Change management reduces accidental outages. After resolution, conduct a post-incident review that identifies root causes and preventive actions rather than blaming individuals.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Network performance and quality of service
Network performance is multidimensional. Bandwidth is the theoretical or configured capacity of a link; throughput is the achieved data rate; latency is the time taken for data to travel; jitter is variation in delay; and packet loss is the proportion of packets that fail to arrive. Real-time voice and video are sensitive to delay and jitter, while bulk file transfer may be more tolerant.
Quality of Service mechanisms classify traffic and apply policies such as prioritisation, shaping, policing and queue management. QoS cannot create additional bandwidth; it allocates scarce resources according to policy. Poorly designed prioritisation can starve other applications. Classification should be based on business needs and monitored for fairness.
Capacity planning uses current measurements, growth projections, peak concurrency and application profiles. A campus may need more upstream capacity during online examinations than during ordinary browsing. Performance investigations should examine the full path, including Wi-Fi, WAN, DNS, server processing, database queries and third-party services.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Data centres and enterprise network design
A data centre houses compute, storage, network and supporting facilities such as power, cooling, physical security and fire suppression. Enterprise network design connects users, branches, data centres and cloud environments while enforcing policy and maintaining availability. Design choices depend on service criticality, regulatory obligations, geography and budget.
Redundancy may be provided through dual power feeds, uninterruptible power supplies, generators, redundant network paths, clustered systems and replicated data. Redundancy is useful only when failure domains are genuinely independent and failover is tested. A shared power supply or configuration error can defeat apparently duplicated components.
Document recovery objectives and dependencies. A network design should identify single points of failure, maintenance windows, capacity limits and out-of-band management. Environmental controls and physical access are part of information security because physical compromise can bypass many logical safeguards.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Cloud computing concepts and service models
Cloud computing provides on-demand access to configurable computing resources over a network, with rapid provisioning and measured service. Common characteristics include self-service, broad network access, resource pooling, elasticity and metered usage. Cloud services can reduce procurement delays and shift some infrastructure responsibilities to providers, but do not remove customer accountability.
Infrastructure as a Service (IaaS) supplies virtualised compute, storage and networking. Platform as a Service (PaaS) provides a managed runtime or development platform. Software as a Service (SaaS) delivers complete applications to users. Function as a Service or serverless computing runs event-triggered code while the provider manages much of the underlying infrastructure.
The shared-responsibility model divides security and operational duties between provider and customer. The provider may secure physical facilities and foundational infrastructure, while the customer remains responsible for identity, data classification, access policy and application configuration. Exact boundaries differ by service and must be read from provider documentation and contracts.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Cloud deployment models and architecture
Public cloud services are operated by a provider and shared across customers through logical isolation. Private cloud environments are dedicated to one organisation, whether on-premises or hosted. Hybrid cloud combines private and public environments; multicloud uses services from more than one provider. These models create different trade-offs in control, portability, cost and operational complexity.
Cloud architecture commonly uses regions and availability zones. A region is a geographic area containing provider infrastructure; zones are separated locations designed to reduce correlated failures. Applications may distribute workloads across zones for resilience. Cross-region recovery can improve continuity but introduces replication delay, data residency questions and higher cost.
Select deployment models based on workload requirements, sensitivity, latency, legal obligations, integration, skills and total cost. Avoid choosing multicloud merely as a slogan: operating multiple platforms requires expertise, consistent governance, identity integration and monitoring. Cloud architecture should be documented as a set of services, data flows, trust boundaries and failure assumptions.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Virtualisation, containers and orchestration
Virtualisation abstracts physical hardware so multiple virtual machines can run on a host. A hypervisor manages virtual machine resources and isolation. Virtual machines include their own operating system environment, which can provide strong separation but uses more resources than some alternatives. Virtualisation supports consolidation, testing, disaster recovery and flexible provisioning.
Containers package an application with its dependencies while sharing the host operating-system kernel. They are generally lighter than full virtual machines and support repeatable deployment. Container images should be minimal, scanned, signed or otherwise verified where appropriate, and built from trusted sources. Secrets should be supplied at runtime rather than baked into images.
Container orchestration platforms schedule workloads, manage service discovery, health checks, scaling and rolling updates. Orchestration adds complexity and requires skills in networking, storage, security and operations. Small systems may be better served by managed application platforms. Use orchestration when its operational benefits justify its learning and maintenance cost.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Cloud storage, databases and data services
Cloud storage includes object, block and file services. Object storage is suited to documents, media, backups and data lakes; block storage provides virtual disks for compute workloads; and file storage exposes shared file-system semantics. Each has different performance, access, durability and cost characteristics. Lifecycle policies can move older data to lower-cost tiers.
Managed relational databases provide structured tables, transactions and query languages, while managed NoSQL services support document, key-value, column-family or graph models. Managed operation can reduce routine maintenance, but customers must still design schemas, access policies, backup strategy, retention and performance. Understand replication, consistency and recovery behaviour before selecting a service.
Storage security includes encryption, access policies, private endpoints where appropriate, versioning, audit logs and deletion controls. Public exposure of a storage bucket or overly broad database credentials can disclose large datasets. Backups should be protected from production compromise and recovery should be tested regularly.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Cloud networking and hybrid connectivity
Cloud networks provide virtual networks, subnets, routing tables, security groups, network access controls and gateways. These constructs map familiar networking concepts into provider-managed infrastructure. Address plans should avoid overlap with on-premises sites and partner networks, particularly where VPN or private connectivity is required.
Hybrid connectivity may use encrypted site-to-site VPNs, dedicated circuits or provider interconnects. The design must consider bandwidth, latency, redundancy, route advertisement, DNS resolution, identity integration and failure behaviour. Private connectivity does not automatically encrypt all traffic or authorise every workload.
Cloud network misconfiguration can expose management ports, databases or storage services to the public Internet. Use infrastructure as code, policy review and automated configuration checks. Apply least privilege to security groups and routing. Maintain diagrams and test that only intended traffic crosses each trust boundary.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Cloud identity, governance and security
Cloud identity and access management controls who can administer resources and what workloads can do. Policies may attach to users, groups, roles and service identities. Prefer short-lived credentials and workload identities over embedded long-lived keys. Separate production and development accounts or projects to reduce accidental access and blast radius.
Governance defines account structure, naming, tagging, approved regions, data classifications, budgets, logging, security baselines and exception processes. Guardrails can prevent risky configurations before deployment. Centralised audit logs help detect suspicious activity, while alerting should be tuned to actionable events and protected against tampering.
Cloud security reviews should assess public exposure, identity privilege, encryption, key management, vulnerability management, backup protection, network paths and incident response. Provider compliance reports may support assurance but do not prove that a customer's specific configuration is compliant. Evidence must be mapped to actual controls and responsibilities.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Cloud economics, cost control and vendor management
Cloud pricing can include compute time, storage capacity, requests, data transfer, managed services, licences and support. Elasticity can improve utilisation, but poorly controlled resources can continue generating charges after a project ends. Cost estimates should include normal, peak, backup, test and data-egress scenarios.
FinOps brings financial accountability into cloud usage through visibility, optimisation and shared responsibility. Tags and account structures help allocate spend to teams and products. Budgets, alerts, rightsizing, scheduling non-production systems and storage lifecycle policies can reduce waste. Cost optimisation should not undermine availability, security or recovery objectives.
Vendor assessment should consider service-level commitments, data location, portability, exit costs, support, incident notification, subcontractors, audit rights and continuity. Avoid assuming that a cloud provider's headline availability figure guarantees end-to-end application availability. Contracts and architecture must address the organisation's own responsibilities.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Availability, backup and disaster recovery
Availability is the proportion of time a service is usable. Reliability concerns the likelihood that it performs correctly over a period. Resilience is the ability to withstand or recover from disruption. High availability uses design and operational controls to reduce downtime; disaster recovery restores service after a major incident.
Recovery Time Objective (RTO) is the target maximum time to restore a service after disruption. Recovery Point Objective (RPO) is the target maximum amount of data loss measured in time. A service with an RPO of 15 minutes aims to recover to a point no more than about 15 minutes before the incident, subject to the chosen recovery design.
Backups should be scheduled, encrypted, access-controlled, retained according to policy and isolated from common attack paths. Replication is not a substitute for backup because corruption or deletion may replicate. Test restoration, failover, communication and business workarounds. A recovery plan that has never been exercised is an assumption, not proven capability.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Network and cloud governance, law and ethics
Network and cloud decisions affect privacy, confidentiality, sovereignty, accessibility, intellectual property and service continuity. Organisations should identify applicable data-protection law, sector requirements, contractual duties and cross-border transfer restrictions. The precise rules depend on the people, data, provider, location and service involved.
Policies should define acceptable use, remote access, network monitoring, retention, incident reporting, privileged administration and cloud procurement. Monitoring should be proportionate, transparent where required and restricted to legitimate purposes. Staff need clear guidance about personal devices, public Wi-Fi, removable media and approved storage.
Ethical infrastructure management considers who may be excluded by unreliable connectivity, unaffordable data or inaccessible interfaces. It also considers energy use, electronic waste, supplier practices and the environmental cost of data centres. Decisions should be evidence-based and reviewed as usage and technology change.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Network and cloud project implementation
A network or cloud project begins with requirements discovery: users, applications, data sensitivity, traffic patterns, locations, availability, compliance and budget. A baseline survey documents current assets, topology, addressing, service dependencies and operational pain points. Requirements should be measurable and prioritised.
Design alternatives should be compared against security, performance, resilience, scalability, skills, cost and exit strategy. A pilot reduces uncertainty before a broad rollout. Migration plans define sequencing, data transfer, DNS changes, testing, communication, rollback and support. Change windows should be selected around business impact and recovery capacity.
After implementation, verify configuration, access controls, performance, monitoring, backup and documentation. Train operators and users. Track defects and service outcomes. A project is not complete merely because infrastructure has been provisioned; it is complete when the intended business service works reliably and ownership has transferred.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Emerging networking and cloud technologies
Software-defined networking separates control logic from forwarding devices and enables programmable network policy. Network function virtualisation implements functions such as firewalls or load balancers as software. These approaches can increase flexibility but require secure controllers, reliable automation and careful failure-domain design.
Edge computing places selected processing close to users or devices to reduce latency, support local operation or limit data transfer. It is relevant to industrial systems, connected transport, retail and remote locations. Edge deployments create distributed patching, physical security, inventory and connectivity challenges.
Modern networking also includes 5G, satellite broadband, intent-based operations, confidential computing and AI-assisted observability. New technology should be evaluated through pilots, threat models, interoperability tests, lifecycle costs and measurable user benefit. Hype is not a substitute for a clear service requirement.
For an operational design, document the requirement, selected approach, assumptions, dependencies and measurable acceptance criteria. Identify what can fail, how failure will be detected, who is responsible for response and how service will be restored. Validate the design through a controlled test and retain evidence so that future administrators can maintain it consistently.
Case study: university campus network redesign
A university has lecture halls, laboratories, libraries, residences, administration offices and remote campuses. Users report slow Wi-Fi during examinations, intermittent access to the learning management system and difficulty connecting visiting researchers. The institution also needs to isolate guest devices and protect student records.
Start by mapping buildings, fibre routes, switch closets, wireless coverage, Internet links, application dependencies and user concurrency. Separate student, staff, guest, research, server and management traffic through controlled segments. Design resilient core links, appropriate access-point density, central identity, secure guest onboarding and monitored DNS and DHCP services.
Test roaming, peak concurrent examination sessions, failover, guest isolation, access to authorised services, backup restoration and incident escalation. Measure latency, packet loss, wireless signal quality, application response time and help-desk incidents. Plan phased migration by building with rollback and user communication.
Develop a topology diagram, asset inventory, traffic-flow matrix, security-zone policy, capacity estimate and implementation schedule for this scenario. Include operational training, monitoring, change approval, user communication and a documented rollback plan. Evaluate the result against user experience, risk reduction, resilience and total lifecycle cost.
Case study: moving a student information system to cloud
A college operates an on-premises student information system that experiences seasonal demand peaks during registration and examination results. Management is considering cloud hosting to improve resilience and reduce hardware maintenance. The database contains personal and academic records subject to legal and institutional controls.
Classify data and determine applicable location and transfer requirements. Compare IaaS, PaaS and SaaS options, considering application compatibility, identity integration, database migration, network connectivity, backup, recovery objectives and exit costs. Use a pilot environment with masked data and automated deployment before migrating production records.
Test data integrity, application permissions, transaction performance, peak registration load, disaster recovery, audit logging and integration with finance and learning platforms. Establish cost budgets, service ownership, support escalation and a tested rollback or recovery plan. Cloud adoption is successful only when service outcomes and governance are demonstrably improved.
Develop a topology diagram, asset inventory, traffic-flow matrix, security-zone policy, capacity estimate and implementation schedule for this scenario. Include operational training, monitoring, change approval, user communication and a documented rollback plan. Evaluate the result against user experience, risk reduction, resilience and total lifecycle cost.
Case study: secure hybrid connectivity for a hospital group
A hospital group operates several clinics, a central data centre and cloud-hosted appointment services. Clinicians need secure access to approved records across locations, while visitors require Internet access. Outages could interrupt appointments and operational coordination.
Design encrypted site-to-site connectivity with redundant paths, segmented clinical, administration, guest and management networks, strong identity and role-based access. Ensure critical local workflows can continue during WAN disruption where safe. Protect endpoints and network devices, centralise security logging and restrict vendor remote access.
Exercise link failure, identity-provider outage, ransomware containment, lost device, backup restoration and clinic isolation. Confirm that guest access cannot reach clinical systems, privileged actions are audited and recovery procedures meet agreed RTO and RPO. Review privacy and clinical safety implications with responsible professionals.
Develop a topology diagram, asset inventory, traffic-flow matrix, security-zone policy, capacity estimate and implementation schedule for this scenario. Include operational training, monitoring, change approval, user communication and a documented rollback plan. Evaluate the result against user experience, risk reduction, resilience and total lifecycle cost.
Network and cloud architecture diagrams
Layered communication model
Application — web, email, DNS and business services
↓
Transport — TCP or UDP process communication
↓
Internet — IP addressing and routing
↓
Link — Ethernet, Wi-Fi and local delivery
↓
Physical — fibre, copper and radio signals
Campus network segmentation
Internet / provider edge
↓
Firewall and redundant core routing
↓
Staff VLAN · Student VLAN · Guest VLAN
↓
Server / data-centre VLAN · Management VLAN
↓
Policy-controlled access to applications and data
Cloud shared responsibility
Cloud provider: facilities, physical infrastructure and foundational services
↓
Shared: platform configuration, identity integration and incident coordination
↓
Customer: data classification, access policy, application code and user governance
↓
Verify responsibilities against each specific cloud service contract
Cloud disaster recovery sequence
Detect and declare incident
↓
Assess service and data impact
↓
Activate recovery environment
↓
Restore data to agreed recovery point
↓
Validate application and identity dependencies
↓
Communicate recovery and resume normal operations
University examination and revision questions
Use labelled diagrams and relevant examples. Model answers identify the central concepts expected in a structured response.
Question 1. Define a computer network and state four organisational benefits.
Model answer: A network connects devices to exchange data and share services. Benefits include communication, resource sharing, centralised services, collaboration, remote access and integrated business processes.
Question 2. Differentiate LAN and WAN.
Model answer: A LAN connects devices within a limited site such as an office or campus; a WAN connects networks across larger geographic distances.
Question 3. Explain star and mesh topologies.
Model answer: Star connects endpoints to a central device; mesh provides multiple paths between nodes. Star is simpler, while mesh can improve resilience at additional cost.
Question 4. State the seven OSI layers in order.
Model answer: Physical, Data Link, Network, Transport, Session, Presentation and Application.
Question 5. What is the purpose of a network switch?
Model answer: It forwards frames within a LAN using learned MAC-address information and supports efficient device connectivity.
Question 6. Explain IPv4 subnetting.
Model answer: Subnetting divides an IP address block into smaller logical networks using a prefix length to identify network and host bits.
Question 7. What is longest-prefix matching?
Model answer: A router selects the most specific matching route for a destination address.
Question 8. Compare TCP and UDP.
Model answer: TCP provides connection-oriented reliable ordered byte-stream delivery; UDP provides connectionless datagrams without built-in delivery or ordering guarantees.
Question 9. State three factors affecting Wi-Fi performance.
Model answer: Signal strength, interference, channel contention, client capability, access-point density and backhaul capacity.
Question 10. Explain the functions of DNS, DHCP and NTP.
Model answer: DNS resolves names, DHCP supplies network configuration dynamically, and NTP synchronises system clocks.
Question 11. What is network segmentation?
Model answer: Dividing a network into controlled zones to limit broadcast scope, separate trust levels and restrict lateral access.
Question 12. Differentiate a firewall and VPN.
Model answer: A firewall enforces traffic policy; a VPN creates an encrypted tunnel across an untrusted network.
Question 13. Define bandwidth, throughput and latency.
Model answer: Bandwidth is link capacity, throughput is achieved data rate, and latency is communication delay.
Question 14. What is Quality of Service?
Model answer: QoS classifies and manages traffic to allocate network resources according to service requirements.
Question 15. Explain the cloud shared-responsibility model.
Model answer: The provider and customer divide security and operational duties; exact responsibilities depend on the service model and configuration.
Question 16. Compare IaaS, PaaS and SaaS.
Model answer: IaaS supplies infrastructure resources, PaaS supplies a managed application platform, and SaaS delivers a complete application.
Question 17. Differentiate public, private and hybrid cloud.
Model answer: Public cloud is provider-operated shared infrastructure, private cloud is dedicated to one organisation, and hybrid combines environments.
Question 18. What is virtualisation?
Model answer: Abstraction of physical computing resources to run multiple isolated virtual environments on shared hardware.
Question 19. Compare containers and virtual machines.
Model answer: Containers share the host kernel and package application dependencies; virtual machines run separate guest operating systems on a hypervisor.
Question 20. Name three cloud storage types.
Model answer: Object, block and file storage, each offering different access and workload characteristics.
Question 21. Why is cloud identity management important?
Model answer: It controls who can access resources and what actions users and workloads may perform, limiting unauthorised activity.
Question 22. Define RTO and RPO.
Model answer: RTO is the target time to restore a service; RPO is the target maximum data loss measured in time.
Question 23. Why is replication not a complete backup strategy?
Model answer: Deletion, corruption or ransomware may replicate to the secondary copy; independent protected backups are needed.
Question 24. State four network troubleshooting steps.
Model answer: Define the symptom, establish scope, test connectivity and configuration, isolate the cause, apply a controlled fix and verify recovery.
Question 25. What is a zero-trust security approach?
Model answer: It avoids implicit trust based on network location and evaluates identity, device, context and resource access continuously.
Question 26. Explain cloud elasticity.
Model answer: The ability to provision or release resources in response to changing demand.
Question 27. What is FinOps?
Model answer: A collaborative practice that improves visibility, accountability and optimisation of cloud expenditure.
Question 28. Why should organisations avoid overlapping IP ranges in hybrid networks?
Model answer: Overlapping ranges complicate routing, VPN connectivity and communication between sites.
Question 29. Describe a cloud migration pilot.
Model answer: A limited controlled migration used to test compatibility, security, performance, cost and operational processes before wider rollout.
Question 30. Give four criteria for selecting a cloud provider.
Model answer: Security, availability, data location, compliance, service capability, support, interoperability, cost and exit arrangements.
Official Education References
Use these official organisations for curriculum guidance, examinations and further learning information.