High Availability & DR Questions
Part of the Cloud Computing Roadmap
In short
"How would you get 99.99% uptime?" and "what if the whole region goes down?" sound similar but test different things. One is about surviving small failures. The other is about recovering from big ones.
'How would you design for 99.99% uptime?' or 'walk me through your disaster recovery plan for X' — testing Phase 11's HA patterns and Phase 13's DR strategy knowledge together.
Think of it like this
Two pens and a plan B
In board exams you carry two pens, so if one dies you keep writing without losing a minute. That's high availability. But if the exam centre floods, you need a different plan: another centre, a re-exam date. That's disaster recovery.
Watch it happen
DR strategies, from cheapest to fastest recovery
1/6Four common disaster recovery strategies. Moving right means faster recovery and a higher monthly bill.
The big ideas
What you need to know
HA: survive small failures
Multiple instances across availability zones, health checks, load balancing and Multi-AZ databases.
DR: recover from big ones
A plan for losing a region, bad data or ransomware. Usually involves another region and tested backups.
RTO and RPO
RTO: maximum time to be back up. RPO: maximum data you can lose, measured in time. Always ask for both.
Untested DR isn't DR
Mention restore drills and failover tests. A backup nobody has restored is only a hope.
See it in code
Walk through the code
1Availability Downtime per year Per 30-day month299% ~3.65 days ~7.2 hours399.9% ~8.8 hours ~43 minutes499.99% ~53 minutes ~4.3 minutes5 6Two parts in series (both must work):7 99.9% x 99.9% = ~99.8%8Two copies in parallel (either can serve):9 1 - (0.001 x 0.001) = 99.9999%, if failures are independent1/3Each extra nine cuts allowed downtime by ten times. 99.99% leaves about four minutes a month.
Know these numbers cold. The series vs parallel maths explains why redundancy helps and why long chains of dependencies hurt.
Save yourself the pain
Common mistakes
Oops
Answering "99.99%" with "I'll use multiple regions" and stopping there.
Do this instead
Cover zones, health checks, database failover, deploy safety and monitoring. Regions are one piece.
Oops
Treating backups as DR without mentioning restore tests or RTO and RPO.
Do this instead
Ask for RTO and RPO, pick a strategy that meets them, and say how you'd test it.
Your mission
Answer two questions out loud
Quick check
A business says it can lose at most 5 minutes of data. Which number is this?
Go deeper
- 1Review Phase 11's HA patterns and Phase 13's DR strategies before this type of interview.
- 2Prepare to discuss RTO/RPO trade-offs for a specific hypothetical scenario.
- 3Practice explaining why multi-AZ differs from multi-region, and when each is justified.
More architecture & design at this stop
Cloud Architecture Design Questions
Cost Optimization Scenarios
Stuck on this topic? Ask an Insider
Get 1:1 guidance from people who've walked this exact path — on the InsideEdge app.
