Imagine posting a jigsaw puzzle split across twenty small parcels — some routed via Edinburgh, others via Birmingham, whichever is fastest. The recipient reassembles them on arrival. That is packet switching: breaking data into small chunks, sending each by the best available route, and reassembling them at the destination.

What is a packet?

A packet is a small chunk of data — typically a few hundred to a few thousand bytes — into which large messages are divided before transmission across a network. Every packet contains two main sections:

Section Contents
Header Source IP address, destination IP address, packet number, total number of packets, checksum (error-detection code)
Payload (data) The actual fragment of the original message (e.g. part of a file, video, or web page)

Some packet formats also include a trailer at the end, containing additional error-checking information.

The header is like the address label and postage on a parcel; the payload is whatever is inside the box.

How does packet switching work step by step?

  1. Splitting. The sending device divides the data into packets, numbering each one so the recipient knows the order.

  2. Routing. Each packet is sent independently across the network. Routers read the destination IP address in the header and forward the packet along the best available path at that moment. Different packets from the same message may travel via completely different routes.

  3. Queuing. If a router is busy, it holds packets briefly in a queue, then forwards them when capacity is available.

  4. Reassembly. The receiving device collects all packets, checks the packet numbers, and reassembles them into the original message in the correct order — even if they arrived out of sequence.

  5. Error checking. The receiving device uses the checksum in each header to verify that the data arrived undamaged. If a packet is corrupt or missing, the receiving device can request that it be resent.

Why does the internet use packet switching rather than a dedicated connection?

Before packet switching, telephone networks used circuit switching: a dedicated physical connection was reserved between caller and receiver for the entire duration of the call. That connection sat idle whenever nobody was speaking — wasting bandwidth.

Feature Circuit switching Packet switching
Dedicated connection? Yes — reserved for the whole call No — shared infrastructure
Bandwidth wasted when idle? Yes No — others use the capacity
Failure tolerance Poor — one break kills the call High — packets re-route around failures
Latency Very low and consistent Variable (depends on congestion)
Best for Real-time voice calls (old PSTN) Data, web, email, streaming

Packet switching lets millions of users share the same physical cables simultaneously, which is why the internet can scale to billions of devices.

What route do packets take?

Packets do not follow a fixed path. Each router makes an independent decision about where to forward the packet next, based on its routing table — a map of the network that is constantly updated. If one link is congested or broken, the router automatically chooses an alternative path.

This dynamic routing is why the internet is described as resilient: losing any single router or cable rarely prevents data from getting through — packets simply take a different route.

What protocols govern packet switching?

Two core protocols work together to deliver packets correctly:

Protocol Role
IP (Internet Protocol) Assigns addresses; defines how packets are addressed and routed between networks
TCP (Transmission Control Protocol) Ensures reliable delivery; numbers packets, detects lost packets, and requests retransmission

Together they form TCP/IP — the foundation of internet communication. UDP (User Datagram Protocol) is an alternative to TCP that skips the reliability checks in exchange for speed, used in live video calls and online games where a slightly dropped packet is better than the delay of a retransmission.

What happens if a packet is lost or arrives out of order?

With TCP: the receiving device acknowledges each packet. If an acknowledgement does not arrive within a timeout period, the sender retransmits the missing packet. The receiving device buffers packets that arrive out of order and re-sequences them before passing the complete message to the application.

With UDP: lost packets are simply ignored. The application receives whatever arrived; if frames are missing in a video call, the picture glitches briefly but playback continues.

Frequently asked questions

What is the difference between a packet and a frame?

A packet is the unit of data at Layer 3 (Network layer) — it contains IP addresses. A frame is the unit at Layer 2 (Data Link layer) — it wraps the packet and adds MAC addresses for delivery across a single network segment. As a packet travels across the internet, it is re-wrapped in different frames at each hop.

How small are packets in practice?

The maximum size is defined by the Maximum Transmission Unit (MTU) of the network, typically 1,500 bytes for Ethernet. A 10 MB photo would be split into roughly 6,700 packets. Large files, therefore, involve thousands of packets travelling independently and being reassembled.

Why might some packets from the same message arrive out of order?

Because different packets may take different routes through the internet, some routes are shorter or less congested than others. A packet sent later might arrive first if it travelled a quicker path. TCP handles reassembly so the application always receives data in the correct sequence.

Do I need to know packet switching for the GCSE computer science exam?

Yes. AQA and OCR both include packet switching as a required topic. You should be able to explain what a packet is (header + payload), describe how packets are routed independently, explain why packet switching is used on the internet, and state an advantage over circuit switching.


Want to test your understanding of how data travels across networks? Visit aitutors.me — Professor Turing will guide you through packet switching questions at exam pace.