Regions & Multi-Region Design
Region pairs · Nonpaired regions · Active-active vs active-passive
Practice questions for this topic →
Region pairs
| Fact | Detail |
|---|---|
| Benefits | Sequential (staggered) platform updates, one region of each pair prioritized for recovery, data residency within the same geography |
| Used by | Some services (for example GRS storage, geo-redundant backups, Backup Cross Region Restore) |
| Not | Automatic failover or automatic resilience. Deploying to a paired region doesn't make an app resilient; Microsoft-managed GRS failover happens only in catastrophic situations |
| Nonpaired regions | Many regions have no pair and rely on availability zones; many services still support geo-replication or multi-region designs to a region you choose |
Multi-region patterns (design guidance)
| Pattern | RTO / cost |
|---|---|
| Active-active | Lowest RTO, highest cost; needs global routing and multi-write data |
| Active-passive (warm standby) | Minutes of RTO; a scaled-down secondary |
| Active-passive (cold / backup and restore) | Hours of RTO; cheapest |
| Redeploy from templates + restore data | Slowest; lowest standing cost |
Zones answer 'a datacenter fails'; regions answer 'the region is gone'. Match the pattern to the stated RTO, RPO and budget.
Exam Tips
Tick each tip once you can answer it without looking.
“Deploying to a paired region makes the app highly available automatically?” → False: you must design your own HA/DR.
“Data must stay within the country/geography during DR” → Choose a secondary in the same geography (often the pair).
“Lowest RTO for a global app, cost secondary” → Active-active multi-region with Front Door + multi-region data (Cosmos multi-write / SQL failover group).
“Minimize cost; RTO of several hours OK” → Backup + restore/redeploy in secondary region (cold standby).
“Region has no pair” → Use availability zones in-region and replicate to another chosen region.
“What region pairs provide” → Sequential updates, prioritized recovery for one region, data residency. Trap: No automatic failover.
“Multi-region patterns by cost” → Active-active (lowest RTO, highest cost), warm standby, cold standby, redeploy. Trap: Pick by the stated RTO, RPO and budget.