Internet and Networking¶

Exam-ready answer

Computer networks are classified partly by geographical scope and administrative ownership. A LAN (Local Area Network) connects devices within a limited site such as a room, building or campus. A MAN (Metropolitan Area Network) interconnects sites across a town or city. A WAN (Wide Area Network) connects networks over regions, countries or continents, normally using carrier infrastructure and routed links.

Example: PCs and printers inside one office form a LAN; connecting several city exchanges creates a MAN; routing traffic from that city to international networks uses a WAN. A PAN is still smaller, such as Bluetooth between a phone and headset.

LANs generally permit high capacity and direct local administration, while MAN/WAN designs require routing, service-level agreements, redundancy and stronger encryption across untrusted or shared facilities. These categories are descriptive, not fixed speed classes: a modern fiber WAN may be faster than an old LAN, and a campus network may span several kilometers. Each site should also have a bounded failure/security domain through VLANs and routers so that one broadcast storm, cable break or compromise does not affect the entire wide-area service.

Practice target: 8–9 minutes; define all three, reproduce five comparison bases, and finish with one linked office-city-country example.

Exam-ready answer

A network topology is the physical arrangement of nodes and links or the logical path by which frames/signals travel. Physical and logical topology can differ; for example, switched Ethernet is physically a star and logically uses point-to-point links.

Example: a modern office normally uses a switched star: disconnecting one PC cable affects only that PC, while redundant core switches and uplinks remove the central single point of failure. A WAN backbone often uses partial mesh because full mesh grows quadratically. Topology alone does not guarantee reliability; loops in Ethernet require spanning-tree or loop-prevention control, ring recovery needs protection switching, and shared bus traffic requires access control. Segmentation, redundant power/links and monitoring should align failure and security domains with service requirements.

Practice target: 8–9 minutes; draw/reference all four arrangements and state exactly what a link, backbone or central-node failure affects.

5-mark answer and 10-mark extension

For 5 marks — write this¶

An IP address is a logical network-layer identifier used by routers to deliver packets to an interface. IPv4 has 32 bits, normally written as four decimal octets, for example 192.168.10.25. Historically, IPv4 used classful networks:

127.0.0.0/8 is loopback and zero-valued ranges have special uses. Classful addressing is obsolete for allocation and routing; modern networks use CIDR notation address/prefix-length. A /n prefix identifies the first \(n\) network bits, and a traditional subnet has \(2^{32-n}-2\) usable host addresses, with /31 point-to-point and /32 host-route exceptions. Private space includes 10.0.0.0/8, 172.16.0.0/12 and 192.168.0.0/16.

DNS maps names to records such as A/AAAA addresses; a URL identifies a resource; HTTP/HTTPS transfers web requests and responses; and FTP supports remote file transfer.

Add for a 10-mark variant¶

These services operate in a layered protocol stack. Application data is carried by TCP or UDP/QUIC, placed in an IP packet, framed by Ethernet/Wi-Fi and transmitted as bits. At the receiver, each layer removes its header and delivers the payload upward.

DNS resolution: the host's stub resolver sends one recursive query to its configured recursive resolver. If the answer is not cached, that resolver makes iterative queries: a root server refers it to the appropriate top-level-domain (TLD) server; the TLD server refers it to the domain's authoritative server; and the authoritative server returns the requested A, AAAA or other record. The resolver caches each response for its TTL and returns the final answer to the client. Root and TLD servers normally give referrals, not the host's final address.

A URL has the general field format

scheme://host:port/path?query#fragment
https://www.example.com:443/docs/page?unit=ict#dns

The scheme selects the protocol, DNS resolves the host, the optional port selects the server process, the path identifies a resource, the query supplies parameters and the fragment identifies a client-side section.

HTTP is a stateless application request/response protocol. A client sends a method, target, headers and optional body; the server returns a status code, headers and body. HTTPS adds TLS authentication, confidentiality and integrity. Typical methods are GET and POST, with status codes such as 200, 404 and 500.

FTP uses a control connection, conventionally TCP port 21, plus a separate active or passive data connection for listings and file contents; old active-mode data commonly originates from port 20. Basic FTP exposes credentials/data, so FTPS adds TLS. SFTP is a different protocol over SSH, normally port 22, not "secure FTP" mode of FTP.

Example: opening the URL above causes DNS resolution, a transport/TLS connection, an HTTP GET and a response; downloading through FTP instead uses its control/data channels. Limitations include finite IPv4 space/NAT, DNS cache poisoning, clear-text HTTP/FTP and server spoofing. CIDR, DNSSEC validation where deployed, HTTPS certificate verification, SFTP/FTPS, firewalls and least-privilege service accounts are appropriate controls.

Practice target: 9 minutes for the five-mark core or 18 minutes for the full answer; label classful addressing as historical and trace DNS through all four resolver/server roles.

Exam-ready answer

A hub is a physical-layer multiport repeater: bits arriving on one port are regenerated to every other port without inspecting addresses. A switch is mainly a data-link-layer bridge: it learns source MAC addresses per port and forwards a unicast Ethernet frame only toward the learned destination, while flooding unknown unicast and broadcast frames within the VLAN. A router is a network-layer device: it examines destination IP addresses, selects a next hop from its routing table, decrements packet lifetime and forwards between different IP networks.

Example: two PCs on the same VLAN exchange frames through a switch without using the default gateway. Traffic to another subnet is sent to the router's MAC address; the router removes the old link frame, routes the IP packet and creates a new frame on the outgoing interface.

A hub permits easy eavesdropping and half-duplex collisions. A switch improves capacity but does not stop a broadcast storm in one VLAN and may face MAC-flooding or loop attacks. A router bounds broadcasts and can apply ACL/firewall/NAT policy, but it adds forwarding/configuration overhead and can be a single point of failure. VLAN separation, port security, loop prevention, authenticated management, ACLs and redundant gateways are common controls.

Practice target: 8–9 minutes; state layer, address used, forwarding action, and exact collision/broadcast-domain behavior for each device.

Exam-ready answer

A server is hardware or, more precisely, a continuously available software process that listens for requests and provides controlled resources to client programs. In the client-server model, a client locates the service, opens a protocol connection or sends a request, the server authenticates/validates it, performs the operation and returns a response.

Process example: a web client resolves the server name, establishes TCP/TLS, sends GET /index.html, and receives an HTTP status, headers and content. Email instead follows client submission → sender SMTP server → recipient SMTP server → mailbox → recipient IMAP/POP client. A print server validates the user, converts or preserves job format, queues it, sends it when the device is ready and retains status/accounting information.

Central servers simplify administration, backup, access control and resource sharing, but create valuable targets and possible availability bottlenecks. Capacity planning, redundancy, patching, TLS, strong authentication, least-privilege service accounts, spam/malware filtering, printer quotas, network segmentation, logs and backups reduce overload, data leakage and service failure. A server protocol must not be exposed more widely than required.

Practice target: 8–9 minutes; define client-server operation and give function, protocol and one transaction for each of the three servers.