Disaster Recovery Strategies#
This is a pure memorization page with a huge payoff: the four DR (disaster recovery) strategies appear in almost every resiliency scenario. Learn them as a cost ↔ recovery time slider.
Two numbers decide everything#
- RPO (Recovery Point Objective) = how much data can I afford to lose? (measured in time back)
- RTO (Recovery Time Objective) = how fast must I be back up? (measured in downtime)
Backup every 24h → RPO = 24h (you lose up to a day). Restore takes 6h → RTO = 6h.
The four strategies#
| Strategy | RPO | RTO | Cost | What’s running |
|---|---|---|---|---|
| Backup & restore | Hours | Hours | $ | Nothing until disaster |
| Pilot light | Minutes | Tens of minutes | $$ | Data services only (DB “always on”) |
| Warm standby | Seconds-minutes | Minutes | $$$ | A scaled-down full copy |
| Multi-site active/active | ~0 | ~0 | $$$$ | Full production in 2+ regions |
flowchart LR
bc[Backup & restore] -->|cheaper, slower| pl[Pilot light]
pl -->|cheaper, slower| ws[Warm standby]
ws -->|cheaper, slower| aa[Multi-site active/active]- Backup & restore - snapshots + AMIs (Amazon Machine Image) + cross-region S3 (Simple Storage Service) replication; “we can afford a day of downtime” scenarios. RDS (Relational Database Service) snapshots, EBS (Elastic Block Store) snapshots, AWS Backup.
- Pilot light - the data layer (RDS Multi-AZ / Aurora) is live in the DR region, but app servers are off. On disaster: start instances from AMIs.
- Warm standby - a smaller but functional environment always running (scaled-down ASG (Auto Scaling Group), read replicas). Scale up on disaster.
- Multi-site active/active - traffic served from both regions simultaneously (Route 53 latency/weighted routing). Zero downtime, double the bill.
Services that make each strategy happen#
- AWS Elastic Disaster Recovery (DRS, ex-CloudEndure) → block-level replication for lift-and-shift servers, minutes-level RPO/RTO (“migrate and protect physical/vmware servers”)
- AWS Backup → tag-based centralized backup plans with retention
- S3 Cross-Region Replication → RPO near zero for objects
- Aurora Global Database → RPO < 1 second, RTO < 1 minute
- Route 53 failover routing → flip DNS to the healthy region
- RDS automated backups → PITR (point-in-time recovery) to any second in the retention window (up to 35 days)
Not: soruda “dakikalar icinde toparlansin ama normalde ortam kucuk kalsin” = warm standby. “Kabul edilebilir kesinti saatler, butce kisitli” = backup & restore. “Sifir veri kaybi” = aktif-aktif ya da Aurora Global + Route 53.
Exam check#
Q1: A company can tolerate up to 4 hours of downtime and up to 1 hour of data loss, and wants the LOWEST cost. Which DR strategy?
- A) Multi-site active/active
- B) Warm standby
- C) Backup & restore
- D) Pilot light
Cevabı göster
Answer: C. Hours-scale RTO/RPO + lowest cost = backup & restore.
Q2: A system requires an RTO of 5 minutes and an RPO of 1 minute across regions. Which combination?
- A) Nightly EBS snapshots copied cross-region
- B) Aurora Global Database with Route 53 failover
- C) Pilot light with AMIs in a second region
- D) S3 versioning
Cevabı göster
Answer: B. Aurora Global gives RPO < 1s / RTO < 1min; the others are too slow.
Ozet#
- RPO = kac saatlik veri kaybi olur, RTO = ne kadar surede ayaga kalkarim
- Backup & restore = ucuz, saatler; Pilot light = veri canli, uygulama kapali
- Warm standby = kucuk ama calisan kopya, dakikalar; Active/active = sifir kesinti, cok pahali
- Aurora Global = RPO<1s, RTO<1dk; DRS = sunucu blok replikasyonu
- Route 53 failover = bolge degistirmenin DNS yolu