Community resourceWorksheet
OCR H446 2.2.2 Performance modelling
Part 10 of 14 · H446 2.2.2 · Computational methods
Software that works perfectly for one user can collapse under a thousand, which is the case for performance modelling in H446 2.2.2. A ticketing service under rising load supplies real-looking figures, and students read a table where response time climbs sharply while throughput barely moves.
Students will:
- distinguish performance modelling from testing whether features work correctly
- interpret load, response time and throughput figures to infer a bottleneck
- recommend a next investigation or design change justified by the evidence
- specify loads, repeated trials, request mixes, measures and success thresholds for a model of their own
- explain why passing a model does not prove the deployed system will meet its target
Inside: 7 explanation cells, 1 multiple-choice question, 2 fill-in-the-blanks cells and 3 written answers. 24 marks, about 25 to 35 minutes.
Series: H446 2.2.2 · Computational methods, part 10 of 14.
Shared by Coding PathwayVerified teacher
- 13 cells
- About 30 minutes
- CC BY-SA 4.0
- Shared 31 Aug 2026
- Updated 3 Sept 2026
Preview
The whole resource, exactly as a class sees it. Answers and marking are held back.
Performance modelling
A ticketing service works for one user, but thousands may arrive when sales open. Performance modelling represents expected workloads before or during development so designers can predict behaviour and make scaling decisions.
By the end, you will be able to
- explain the purpose and stages of performance modelling;
- distinguish functional correctness from performance under load;
- interpret response-time and throughput evidence;
- design and evaluate a useful model.
Reactivate: a model simplifies reality for a defined purpose and has assumptions.
Vary load, measure behaviour, make a decision
A performance model may simulate or represent concurrent requests, transaction rates, data volumes or component delays. It measures outcomes such as response time, throughput, queue length, CPU use or memory use. Testing should move from ordinary values towards peaks and beyond expected capacity so the point of degradation becomes visible.
This is not merely checking for program bugs. A system can return correct answers but too slowly when load increases.
Worked interpretation
| concurrent users | mean response | completed requests/s |
|---|---|---|
| 200 | 0.4 s | 190 |
| 600 | 0.8 s | 540 |
| 1000 | 2.7 s | 610 |
| 1400 | 7.9 s | 620 |
From 1000 users, response time rises sharply while throughput barely increases. This suggests a bottleneck near 600–620 requests/s and growing queues. The useful decision is not ‘the model is slow’; it is to locate the constrained component, test a change and compare the measures again.
Which activity best represents performance modelling?
- AChecking whether the login button has the correct colour
- BSimulating increasing concurrent bookings and measuring response time and throughput
- CCorrecting a syntax error
- DAsking one user whether the system feels modern
- load
- throughput
- decision
- variable name
Guided evidence reading
Use the table to make one claim in this form: comparison of two loads → numerical change → inferred bottleneck or user consequence → next test. Avoid declaring an exact real-world capacity unless the model’s workload, hardware and assumptions match deployment.
Interpret the table and recommend one next investigation or design change.
Quote at least two values and explain what their relationship suggests.
Students type their answer here.
Independent model design
The service expects a normal 300 users, a launch peak of 1200 and occasional bursts of 2000. Its payment provider adds variable delay. Specify loads and repeated trials, realistic request mixes, measures, assumptions and success thresholds. Include escalation rather than testing only the predicted average.
Design a performance model for the ticketing service and explain how its results would inform release or scaling decisions.
Include independent variable, representative conditions, at least three measures, thresholds and limitations.
Students type their answer here.
Explain why passing the model does not prove the deployed system will always meet its target.
Link at least two assumptions or omissions to possible real behaviour.
Students type their answer here.
Closed-book checkpoint
Complete each sentence from memory. There is no answer bank and correctness is held for teacher review.
Review your understanding
Before submitting, check that you can explain the central distinction in your own words, expose the intermediate state that supports your answer and apply the method in an unfamiliar context.