Practice & Reference · Exam Simulator

Exam Simulator

20 original practice questions, timed at 20 minutes, covering the real domains behind the three certifications on the Certifications page — the Google Cloud Professional Cloud DevOps Engineer, the DevOps Institute's SRE Foundation, and AWS Certified CloudOps Engineer – Associate. These are original questions testing the same underlying knowledge, not reproductions of any vendor's actual exam content.

☺ Explain it like I'm 10

It's a practice fire drill, not the real fire. Nobody grades you and nothing is on your permanent record — you get a clock, twenty questions, and an honest score at the end so you know exactly what to study next.

🦥🐢Your hosts for this topic: Sol the Sloth & Timmy the Turtle — Sol insists on doing the error-budget arithmetic slowly and correctly, Timmy checks every answer before it counts.

Before you start

☺ Like you're 10: One question at a time, twenty minutes on the clock, and you can go back and change an earlier answer before you submit.

20 questions, single-select, one at a time — use Previous and Next to move around before submitting, your answers are kept as you go. The 20-minute timer starts the moment you click start and auto-submits when it hits zero. Your best score is saved on this device only — nothing is sent anywhere.

Q1. What is the precise difference between an SLI and an SLO?

They mean exactly the same thing. An SLI is the actual measured metric; an SLO is the target you set for that metric. An SLO is measured; an SLI is the target. SLIs only apply to databases.

Q2. A service has a 99.9% SLO over a 30-day month. Roughly how much downtime does that allow?

About 43 minutes About 4.3 hours About 8.6 minutes About 3 days

Q3. Per the Google SRE book's definition, which of these is the clearest example of "toil"?

Designing a new service's architecture. Manually restarting the same crashing service every few days with no permanent fix. Writing an automated test suite. Running a chaos engineering game day.

Q4. Which golden signal measures how "full" a service's resources are — CPU, memory, or queue depth near their limit?

Latency Traffic Errors Saturation

Q5. During a live incident, what is the primary job of the Incident Commander?

To personally write the code fix. To coordinate the response and make decisions — not necessarily to be the one debugging. To decide who gets blamed once it's resolved. To handle only the marketing response.

Q6. What is the goal of the "five whys" technique in a postmortem?

To find someone to blame within five questions. To drill past the first, surface-level cause to a deeper, systemic one. To limit the postmortem meeting to five minutes. To identify exactly five action items, no more, no less.

Q7. Why does capacity planning deliberately keep headroom above current baseline load rather than running at exactly current capacity?

Headroom is wasted money and should always be minimized to zero. It absorbs a traffic spike or a partial capacity loss without an immediate overload. It is legally required in every jurisdiction. It has no effect on reliability at all.

Q8. What does a circuit breaker do when a downstream dependency starts failing?

It retries the failing call infinitely until it succeeds. It stops calling the failing dependency for a cooldown period instead of piling up failed or slow calls. It automatically restarts the entire service. It deletes the failing dependency from the codebase.

Q9. Why does adding random "jitter" to an exponential backoff retry matter?

Jitter has no real effect and is purely cosmetic. It prevents many clients from retrying in synchronized waves that could overwhelm a recovering service. It makes each individual retry succeed faster. It's required by the HTTP specification.

Q10. In progressive delivery, what typically triggers an automated rollback?

A calendar reminder. Golden-signal metrics or SLO burn rate for the canary cohort crossing an unhealthy threshold. A developer's personal preference. The end of the business day.

Q11. What makes a chaos engineering experiment "hypothesis-driven"?

You randomly break things with no plan. You form a specific, falsifiable prediction about system behavior under a specific failure, then inject that failure and observe whether it holds. You only run experiments in production with no safety net. You skip forming any prediction and just see what happens.

Q12. What does "blast radius control" mean when starting a chaos experiment?

Starting with the smallest possible scope — one instance, a small percentage of traffic, non-production first — and expanding as confidence grows. Always testing 100% of production traffic immediately. Only testing during a scheduled maintenance window, never otherwise. Measuring the physical size of the data center.

Q13. When a team's error budget is fully exhausted, what does a healthy error-budget policy typically do?

Nothing changes at all. Freeze risky feature releases and prioritize reliability work until the budget recovers. Immediately fire the on-call engineer. Increase the SLO target to make the budget look better.

Q14. What's the difference between MTTR and MTTD?

They're the same metric under two names. MTTD is mean time to detect a failure; MTTR is mean time to restore/resolve it once detected. MTTR only applies to hardware failures. MTTD is measured in dollars, not time.

Q15. Roughly how much downtime per year does a 99.99% ("four nines") availability target allow?

About 52 minutes About 8.7 hours About 3.65 days About 5 minutes

Q16. The bulkhead pattern, named after ship compartmentalization, is used to:

Isolate resource pools per dependency so one slow dependency can't exhaust resources needed by everything else. Merge all dependencies into a single shared connection pool. Physically separate data centers by ocean. Encrypt traffic between two services.

Q17. What is "load shedding"?

Automatically scaling up more servers. Deliberately rejecting some requests under extreme load to keep serving the rest, rather than degrading for everyone. Deleting old log files to free disk space. Reducing the number of on-call engineers.

Q18. What's the difference between RTO and RPO in disaster recovery planning?

RTO is how long recovery may take; RPO is how much data loss (measured in time) is acceptable. They are the exact same measurement. RPO only applies to on-premises systems. RTO is measured in dollars.

Q19. Which single domain carries the highest published weight on both the Google Cloud Professional Cloud DevOps Engineer exam (tied) — CI/CD pipelines and observability/troubleshooting — each at what weight?

12% each 25% each 18% each 50% each

Q20. Which currently-active AWS certification replaced the retired "AWS Certified SysOps Administrator – Associate" exam (last bookable 29 September 2025)?

AWS Certified DevOps Engineer – Professional AWS Certified CloudOps Engineer – Associate (SOA-C03) AWS Certified Security – Specialty No replacement was issued.

⚠ What this is and isn't

A good score here means you know the underlying concepts, not that you're guaranteed to pass a specific vendor's exam. Pair this with real practice: set a real SLO on a toy service and watch its error budget actually burn — see the toolchain and checklist pages.

Exam Simulator

Before you start

Your results