Computing case analysis: evaluate a proposed system
Introducido| English | Significado |
|---|---|
| case analysis/keɪs əˈnæləsɪs/ | Studying a real or realistic business situation to identify its problems and recommend what to do. |
| evidence/ˈevɪdəns/ | Information or observations used to support or challenge a claim. |
| throughput/ˈθruːpʊt/ | rendimiento |
Reused code destroys a rocket, 1996
- On 4 June 1996 the first Ariane 5 rocket veered off course about 37 seconds after launch and was destroyed. The inquiry found that guidance software reused from Ariane 4 converted a 64-bit floating-point velocity value into a 16-bit integer. Ariane 5 flew faster, the value overflowed, and the backup and main systems both shut down.
- The code had worked on Ariane 4. Nobody had re-checked it against the new rocket's requirements. Case analysis 案例分析 means evaluating a proposed system against its context, with evidence 证据.
What caused the Ariane 5 failure in 1996?
Dificultad 1/5 · 100 monedas + un pequeño bono en el primer intento de hoy
The new rocket flew faster.
Start from requirements
- List the stakeholders and what each needs; turn needs into measurable criteria.
- Ask what is different about this context: users, load, data, risks.
Which is a measurable criterion?
Dificultad 1/5 · 100 monedas + un pequeño bono en el primer intento de hoy
Turn needs into numbers.
Evidence, not assertion
- Prototype tests, load measurements, user trials and expert review count as evidence. A vendor's claim does not.
- Compare options on the same criteria.
Which count as evidence?
Dificultad 2/5 · 140 monedas + un pequeño bono en el primer intento de hoy
Claims are not evidence.
Risks and ethics
- Security, privacy, reliability, accessibility, cost and environmental impact. Who could be harmed, and how?
- Plan mitigations: backups, testing, data minimisation.
Worked example: choosing a canteen ordering system
Requirement: serve 1,200 students in a 30-minute lunch — at least 1,200 ÷ 1,800 s ≈ 0.67 orders per second, with headroom for peaks.
- Prototype A: 4 successful requests per second; Prototype B: 5 — in a 45 s trial B completed 225 requests.
- Both exceed the requirement, so throughput does not decide. Other criteria: A works offline when Wi-Fi fails; B stores card numbers on the school server.
- Recommendation: A, because offline operation protects the lunch rush and it avoids storing payment data, a privacy risk. Evidence is measured, not claimed.
A prototype completes 360 successful requests in 45 s. What is its throughput in requests per second?
Dificultad 2/5 · 140 monedas + un pequeño bono en el primer intento de hoy
360 ÷ 45 = 8.
Why does throughput not decide between A and B in the canteen example?
Dificultad 1/5 · 100 monedas + un pequeño bono en el primer intento de hoy
Other criteria decide.
Structuring an evaluation
- Criteria → evidence for each option → weigh trade-offs → justified recommendation → limitations.
- Say what further evidence would change the decision.
A good evaluation states what further evidence could change the decision.
Dificultad 1/5 · 100 monedas + un pequeño bono en el primer intento de hoy
State limitations.
"It worked before" is not evidence for a new context. Ariane 5's software had flown safely on Ariane 4. Re-test reused components against the new requirements.
Code that worked in an old system needs no re-testing in a new one.
Dificultad 1/5 · 100 monedas + un pequeño bono en el primer intento de hoy
Ariane 5 shows why.
Evaluate a proposed system against measurable requirements for its context, using evidence from tests rather than claims. Weigh risks and trade-offs, give a justified recommendation, state its limits, and re-test anything reused.
A school proposes face-recognition attendance. Write two criteria, the evidence you would need for each, one risk and a recommendation.
Criteria, evidence, decision.