A2.2 · Book pp. 127–135
Network architecture
How a network is laid out decides what happens when something breaks. Design it well and a failure goes unnoticed.
What you need to know
A2.2.1 Network topologies
A topology is the layout of nodes and connections in a network.
- Star: all nodes connect to one central device, usually a switch.
- Mesh: every node links directly to the others. A partial mesh links only some of them.
- Hybrid: a mix, such as meshed base stations with a star around each.
| Factor | Star | Mesh | Hybrid |
|---|---|---|---|
| Reliability | Hub is a single point of failure | Very high, with many paths | High. A failure affects only nearby clients |
| Speed | Limited by the central device | High, with direct links | Depends on the design. Watch for bottlenecks |
| Scalability | Limited by ports on the hub | Each new node needs many links | Highly scalable and adaptable |
| Collisions | Minimal, with dedicated links | Minimal, with direct links | Depends on the design |
| Cost | Low at first | High, with so much cabling | Higher to start, efficient long-term |
| Typical use | Homes and small offices | Military, aviation, disaster areas | Large enterprises and telecoms |
links in a full mesh = n(n − 1) ÷ 2
⚠️ Common mistake
Learn to draw each topology, and always say why one suits the situation.
HL A2.2.2 The function of servers
A server provides a service to other devices. Here it mostly means the software. One computer can run many servers at once, each on its own port: web on 80 and 443, FTP on 20 and 21, SSH on 22 and DNS on 53.
| Server | Function | Scalability | Reliability | Security |
|---|---|---|---|---|
| DNS | Translates domain names into IP addresses | Distributed database and caching | Many redundant servers | A frequent target, so it needs firewalls and intrusion detection |
| DHCP | Assigns IP addresses and settings | Address pools can be managed per network | Devices cannot join if it fails | Should only answer on trusted networks |
| File | Stores and shares files centrally through FTP, SFTP, SMB, NFS or AFP | Add drives and faster network cards | RAID can survive a failed drive | Read-only or read-write permissions |
| Stores email and exchanges it with other servers using SMTP and POP3 | Often outsourced to providers | Queues and redundant systems | SPF names approved senders. DKIM adds a digital signature | |
| Proxy | Acts for clients, and caches content | A reverse proxy spreads load across servers | Balancing load gives fault tolerance | Shields internal devices from the internet |
| Web | Hosts pages, either static or generated at runtime | Load spread across servers behind a reverse proxy | Sites go down if it fails | Needs DDoS protection, and TLS for HTTPS |
✍️ Try it
Run nslookup example.com on Windows or host example.com on a Mac. To do the same in Python:
# pip install dnspython
import dns.resolver
domain = "www.example.com"
try:
for record in dns.resolver.resolve(domain, "A"):
print(record.to_text())
except dns.resolver.NXDOMAIN:
print("That domain does not exist.") ✏️ Book correction
The book says dns.resolver is built in, and imports it as dnspython. It is a third-party package: install dnspython with pip, then use import dns.resolver, as above.
Proxy and reverse proxy
A proxy acts for clients, so a school can cache common pages and reduce traffic. A reverse proxy sits in front of servers, spreading requests across them and caching common replies.
A2.2.3 Networking models
In the client-server model, clients request services and servers provide them. In peer-to-peer, each device is both client and server.
| Factor | Client-server | Peer-to-peer |
|---|---|---|
| Control | Centralized, easy to manage and update | No central authority, hard to coordinate |
| Failure | Server is a single point of failure | No single point of failure, but peers may vanish |
| Security | Managed in one place, but a breach exposes everything | Patches and settings are hard to keep in step |
| Cost | Expensive server hardware and software | Cheap, as peers share the load |
| Scaling | Upgrade the server | Each peer adds capacity |
| Examples | Web, email, online banking | BitTorrent, voice calls, blockchain |
A2.2.4 Network segmentation
Segmentation splits a network into smaller, logically separate parts. It improves security, keeps heavy traffic contained and makes management easier.
| Subnetting | VLANs | |
|---|---|---|
| Separates by | Ranges of IP addresses | Logical groups, often using MAC addresses or logins |
| Example | Staff on 192.168.0.x, students on 192.168.1.x | A guest VLAN and a staff VLAN on the same switches |
| Security | A user might change their IP address to move subnet | Membership is controlled, so it is stronger |
| Reach | Shares the same equipment | Can even span several sites, for example over a VPN |
⚠️ Common mistake
Segmentation is the general idea. Subnetting and VLANs are two ways of doing it.
Quick quiz
HL Extra questions
Exam-style questions
These are written for this site in the style of Paper 1. Answer on paper first, then open the mark points.
-
Outline three benefits of a star topology for a small office. [3]
Mark points
- A cable failure affects only one device [1].
- It costs little for a small number of nodes [1].
- Each device has its own link, so there is minimal risk of collisions [1].
-
A rescue service needs a network that keeps working after any one link fails. Justify a mesh topology. [3]
Mark points
- Each node has several paths [1].
- Traffic can be rerouted if one link or node fails [1].
- The extra cost of links is justified because communication is critical [1].
-
Explain one disadvantage of peer-to-peer for online banking. [2]
Mark points
- There is no central authority to keep the records consistent and secure [1].
- So balances could be inconsistent or a peer could be unavailable when needed [1].
-
Compare client-server and peer-to-peer networking. [4]
Mark points
- Client-server has central control, while peers are equal [1].
- A server is a single point of failure, while P2P spreads the risk [1].
- Servers cost more, but P2P has no such hardware cost [1].
- Servers are easier to secure and manage, but P2P capacity grows with each peer [1].
-
Explain how VLANs improve a school network. [3]
Mark points
- Staff and students are placed in separate logical networks on the same hardware [1].
- Traffic is isolated, so students cannot reach staff servers [1].
- Broadcast traffic stays within a VLAN, which reduces congestion [1].
HL Exam-style questions
-
Describe how a client finds the IP address of a website through DNS. [3]
Mark points
- The client sends the domain name to its configured DNS server [1].
- If the server does not know, it asks another server up the chain [1].
- The address is returned to the client and cached for later requests [1].
-
Describe a proxy server and a reverse proxy, and give one benefit of each. [4]
Mark points
- A proxy makes requests on behalf of clients [1].
- It can cache content, which reduces external traffic [1].
- A reverse proxy receives requests on behalf of servers [1].
- It spreads load across servers, which improves scalability [1].