Protocols and Layers
Every protocol named in the specification and what each one is actually for, why TCP and UDP are different jobs rather than rivals, and what layering buys you.
What a protocol is
A protocol is a set of rules governing how devices communicate. Both ends must follow the same rules or nothing can be understood.
That is more than a definition. It is why standards matter: a laptop made in one country can talk to a server made in another because both implement the same agreed rules.
The protocols AQA names
| Protocol | Stands for | What it does |
|---|---|---|
| Ethernet | The family of protocols for wired LANs, defining how frames are sent over cable | |
| Wi-Fi | The family of protocols for wireless LANs | |
| TCP | Transmission Control Protocol | Splits data into packets, checks they all arrive, and reassembles them in order |
| UDP | User Datagram Protocol | Sends packets with no checking and no reassembly, which is faster |
| IP | Internet Protocol | Addresses packets and routes them between networks |
| HTTP | Hypertext Transfer Protocol | Requests and delivers web pages |
| HTTPS | HTTP Secure | The same, but encrypted so it cannot be read if intercepted |
| FTP | File Transfer Protocol | Transfers files between computers |
| SMTP | Simple Mail Transfer Protocol | Sends email |
| IMAP | Internet Message Access Protocol | Retrieves email, leaving it on the server so it syncs across devices |
The single most commonly asked pair on this topic. Sending goes out with SMTP, reading comes back with IMAP.
- SMTP sends email, from the sender to their server and between mail servers.
- POP downloads mail and normally deletes it from the server, so it lives on one device.
- IMAP keeps mail on the server, so a phone and a laptop both show the same mailbox.
TCP and UDP
These are not competitors. They are different tools for different jobs, and a question asking you to choose is really asking whether reliability or speed matters more.
TCP guarantees delivery. It numbers each packet, waits for the receiver to acknowledge each one, resends anything that does not arrive, and puts the packets back into order at the far end.
Use it when every byte must arrive: a web page, a file download, an email, a bank transfer. A missing byte in a downloaded program is a corrupt program.
UDP sends packets and does not check. Nothing is acknowledged, nothing is resent and nothing is reordered.
Use it when speed matters more than completeness: live video, voice calls, online gaming. A resent video frame would arrive after the moment it belonged to had passed, so resending it is worse than useless.
"UDP is used for a live video call because a lost packet would arrive too late to be useful, and waiting for it would add delay to the whole call." That is the mark, not "UDP is faster".
Layers
Network protocols are organised into layers, each of which handles one part of the job and passes its result to the layer below.
A four layer model is used:
| Layer | Job | Protocols |
|---|---|---|
| Application | Provides services to the user's software | HTTP, HTTPS, FTP, SMTP, IMAP |
| Transport | Splits data into packets and manages the connection | TCP, UDP |
| Network | Adds addresses and routes packets between networks | IP |
| Link | Sends the data over the physical medium | Ethernet, Wi-Fi |
Each layer wraps what it receives from the layer above in its own header. The receiving computer unwraps them in the reverse order, so each layer only ever deals with the header its counterpart added.
Why layering is used
It divides a huge problem into manageable parts. Nobody has to understand the whole of networking at once.
Layers can be changed independently. Moving a laptop from Ethernet to Wi-Fi changes the link layer only. HTTP, TCP and IP carry on exactly as before, and no application has to be rewritten.
It allows specialisation and competition. Different manufacturers can build products for different layers, and as long as each follows the standard for its layer, everything interoperates.
It makes faults easier to isolate. If a website loads by its IP address but not by its name, the problem is in name resolution, not in the cable.
They do not. Layering is about manageability, interoperability and independent change, and that is what the mark scheme rewards.
- Your message starts at the top, in the application.
- Transport splits it into segments and adds a port number.
- Internet wraps each segment in a packet and adds IP addresses.
- Link turns the packet into a frame with MAC addresses, and it goes out.
- At the other end the same four layers run in reverse, unwrapping it.
Key terms you must be able to define
- Protocol
- A set of rules governing how devices communicate, which both ends must follow.
- TCP
- Transmission Control Protocol, which splits data into packets, checks delivery and reassembles them in order.
- UDP
- User Datagram Protocol, which sends packets without checking delivery or reassembling them.
- IP
- Internet Protocol, which addresses packets and routes them between networks.
- HTTP
- The protocol used to request and deliver web pages.
- HTTPS
- HTTP with encryption, so intercepted traffic cannot be read.
- SMTP
- The protocol used to send email.
- IMAP
- The protocol used to retrieve email while leaving it on the server.
- Layer
- One level of a networking model, handling one part of the communication task.
- Encapsulation
- Each layer wrapping the data it receives in its own header before passing it on.
Flashcards
Say the definition out loud before you flip the card. Retrieving it from memory is what builds recall, reading it again is not.
Reach the top grade on this topic
This topic rewards knowing why, not just what.
Justify TCP or UDP from the consequence of loss. Ask what happens if a packet goes missing. If the result is corrupt data, TCP. If the result is a moment that has already passed, UDP.
Explain layering with a change. The clearest evidence of understanding is an example: swapping Wi-Fi for Ethernet changes the link layer and nothing above it, so no application needs rewriting.
Give HTTPS its mechanism. Not "it is secure" but "the data is encrypted before transmission, so intercepted traffic cannot be read without the key".
Mistakes that cost marks here
- Swapping SMTP and IMAP. SMTP sends, IMAP retrieves.
- Saying UDP is better than TCP. They serve different purposes, and the right one depends on whether loss or delay is worse.
- Saying layering makes a network faster. It makes it manageable and interoperable.
- Describing HTTPS as a different protocol from HTTP. It is HTTP with encryption added.
- Saying IP guarantees delivery. IP addresses and routes packets; TCP is what checks they arrived.
TCP and IP were originally a single protocol. They were split in 1978 precisely so that applications needing speed more than reliability could use IP without paying for TCP's guarantees, which is where UDP came from.
Knowledge check
Exam-style questions
Five written questions in the style of the real paper. Write a full answer, then let the marker check it against the mark scheme.
- LA set of rules governing how devices communicate
A protocol is a set of rules that governs how devices communicate, which both the sending and the receiving device must follow for the communication to be understood.
- LTCP checks that every packet arrives and reassembles them in order
- LA downloaded file would be corrupt if any part were missing
- LUDP sends packets without checking or resending, so it is faster
- LA resent video packet would arrive too late to be useful, and waiting for it would add delay
TCP numbers every packet, waits for the receiver to acknowledge each one, resends anything that fails to arrive and puts the packets back into the right order at the far end. That is essential for a file download, because a file with a missing or misordered section is corrupt and will not work at all. UDP does none of that checking, which makes it faster and removes the delay that waiting for acknowledgements introduces. For a live video call that trade is the right one: a packet that was lost and then resent would arrive after the fraction of a second it belonged to had already been displayed, so it is useless, and the delay caused by waiting for it would degrade the whole call for everyone.
- LEach layer handles one part of the task, so the problem is broken into manageable parts
- LDevelopers can work on one layer without understanding the whole system
- LA layer can be changed or replaced without affecting the layers above or below
- LGives an example, such as changing from Ethernet to Wi-Fi affecting only the link layer
The first reason is that layering divides a very large problem into parts small enough to handle. Each layer is responsible for one aspect of communication and only has to interact with the layer directly above and below it, so a developer working on an email client does not need to understand how a signal is placed onto a cable. The second is independence. Because each layer only relies on the interface the next one presents, a layer can be replaced entirely without disturbing the others. Moving a laptop from an Ethernet cable to Wi-Fi changes the link layer completely, but IP, TCP and HTTP carry on unchanged, which is why no application has to be rewritten when the connection type changes.
- LHTTPS encrypts the data before it is transmitted
- LIntercepted traffic cannot be read without the key
- LSo personal data such as logins and messages is protected in transit
HTTPS carries the same HTTP requests and responses over an encrypted connection, so the data is scrambled before it leaves the user's device and is only unscrambled at the far end. Anybody who intercepts the traffic in between, for example on a shared public network, sees only unreadable data rather than the actual content. That matters because users log in to the school site and may send personal information through it, and under plain HTTP a username and password would travel as readable text that anyone on the same network could capture.