Network Rate, Latency and Transfer Time
Separate capacity, useful throughput and delay. Convert the payload into consistent units before deciding whether a task is limited by data volume or by waiting for a response.
Before you start
- Convert decimal MB to bits and apply a supplied efficiency factor.
- Distinguish a round trip from a one-way delay and sequential from overlapping work.
Watch, pause, try
Video by Craig'n'Dave. Use the shared AQA explanation for rate, bandwidth and latency distinctions. These Pearson exercises supply their own raw rate, useful-payload fraction and dependent-request model; baud rate or a nominal link label must not silently become the application’s useful throughput.
Viewing focus: Distinguish bandwidth, throughput, latency and range
After watching, attempt these questions before opening the answers. You can also use the written lesson below.
After watching 1: A 7.5 MB decimal payload crosses a nominal 30 Mbit/s link. Assume useful payload throughput is consistently 80% of that rate and ignore setup. Find useful throughput and ideal payload time. What assumption would be wrong if a question already supplied a useful rate?
Show answer and reasoning
The payload is 7,500,000×8=60,000,000 bits. Useful throughput is 0.8×30,000,000=24,000,000 bits/s, so time is 2.5 s. Applying the same efficiency reduction again to a rate already described as useful would double-count that overhead. State what each rate measures before using the transfer equation.
After watching 2: Before that transfer, three small request/response exchanges must complete sequentially, each adding 50 ms round-trip waiting. Ignore their payload and all other work. Find total time. If those three exchanges were independent and could start together, would their waiting contribution necessarily still be 150 ms?
Show answer and reasoning
The stipulated sequential waits add 3×50=150 ms, so total is 2.65 s. Independent exchanges that start together could overlap their waits, giving a 50 ms waiting contribution in the idealised equal-delay model rather than necessarily 150 ms. Real serialization, server work and other timing costs need separate assumptions; do not add or hide them without specifying the workload.
What you'll be able to do
- Distinguish bandwidth, throughput, latency and range
- Calculate transfer times using SI and IEC units
- Explain why a high-rate link can still feel slow for small interactions
Measure different properties separately
Bandwidth or link capacity describes the available transmission resource under the model. Throughput is the achieved transfer rate, while application goodput counts useful application payload rather than every transmitted header or retry. A nominal link figure does not guarantee that one user receives that entire rate continuously across an end-to-end route.
Latency is delay between specified events, such as a one-way message delivery or a request/response round trip. State which is measured; one-way and round-trip figures are not interchangeable. Range describes the distance over which a suitable link can operate under its environmental and service requirements. Increasing a rate number does not automatically extend range or remove propagation and processing delay.
Convert bits, bytes and prefixes
For an ideal transfer at a constant useful rate, time equals payload bits divided by useful bits per second. A byte contains eight bits. Decimal network prefixes use powers of 1000; IEC capacity units such as MiB use powers of 1024. A 3 MiB payload therefore contains 3×2²⁰×8=25,165,824 bits.
At 12 Mbit/s of useful payload rate, that transfer takes 25,165,824/12,000,000=2.097152 seconds. If a question supplies raw link rate rather than useful rate, include its header/encoding/efficiency assumptions before using the equation. A percentage utilisation factor must be applied to the appropriate rate, not silently subtracted as a fixed number of seconds.
Choose the model that fits the interaction
A tiny interactive request can be dominated by setup and round-trip latency even on a high-capacity link. A large continuous transfer can be dominated by sustained useful throughput. If four dependent request/response exchanges each add 40 ms round-trip waiting, they add 160 ms even when their payloads are tiny. Independent exchanges may overlap, so a simple sum needs an explicit dependency assumption.
End-to-end transfer also depends on the slowest relevant path capacity, congestion, server processing and retransmissions. A wired LAN upgrade alone cannot guarantee a faster remote server. Compare the same workload and define where timing begins and ends. Use units in every step so a factor-of-eight or decimal/binary-prefix mistake is visible before interpreting the answer.
| Quantity | Meaning in a calculation |
|---|---|
| 3 MiB | 3×2²⁰ bytes |
| 12 Mbit/s useful rate | 12×10⁶ useful bits each second |
| One-way latency | Delay in one direction |
| Round-trip latency | Delay for an outward/return interaction |
| Range | Permitted link distance under stated conditions |
Work through the problem
Attempt the task before opening its worked response. Record each change of state, explain the rule used, and check the result against the original requirements. Code examples labelled Python 3 use that language’s conventions; pseudocode states its own assumptions.
Example 1Transfer 3 MiB at a constant useful rate of 12 Mbit/s, ignoring all other time.
Example 2Four small exchanges must happen sequentially; each incurs a 40 ms round trip. What minimum waiting contribution does this model give?
Equation recap
Common mistakes to avoid
Key takeaways
- Rate, achieved throughput, latency and range are distinct.
- Use consistent bit units and stated prefixes.
- Choose a timing model for the actual workload.
Explore further
- Pearson YCP01 official specification (opens in a new tab)
Check the International A-level requirements and relevant numbered points; this course remains in progress.
- Pearson International Computer Science subject guide (opens in a new tab)
Official introduction to the four examined units and their theory/practical purposes.
- Topic reading and reference (opens in a new tab)
Use the relevant board requirement or primary technical reference alongside the written examples.
- 134. AQA A Level (7516-7517) SLR20 - 4.9.1 Bit rate, baud rate, bandwidth and latency — Craig'n'Dave (opens in a new tab)
Original publisher or creator source for the embedded explanation; apply the stated viewing scope.
Return to this tomorrow
Close the notes and answer from memory. Revisit any explanation you cannot yet reconstruct.
- A high-capacity connection guarantees zero latency:
- Before dividing payload bytes by bit/s, convert by:
Test yourself
Answer a few questions, then review anything you missed.
Keep practising
Start with the quiz above to check the key ideas. You can also choose another topic from the practice library.