Design business continuity solutions
Define the targets first
RTO is the maximum acceptable downtime; RPO is the maximum acceptable data loss measured in time. Every architecture question is really asking which pattern meets the stated RTO/RPO at the lowest cost. Backup and disaster recovery are different products: backup protects data over time against deletion, corruption, and ransomware; Site Recovery replicates a workload and orchestrates failover and failback.
Compute and platform patterns
Active-active gives near-zero RTO at the highest cost and is only viable for workloads that can run multi-writer. Active-passive warm standby keeps a scaled-down copy running. Pilot light keeps only core components alive and scales out on failover. Backup/restore is the cheapest and slowest. For availability within a region, availability zones (separate datacenters, 99.99% target) beat availability sets (fault and update domains inside one datacenter, 99.95%).
Data protection specifics
Recovery Services vault handles IaaS VM backup, SQL/SAP HANA in VMs, Azure Files, and the MARS agent; Backup vault handles newer workloads such as blobs, managed disks, and Azure Database for PostgreSQL. Cross-region restore requires geo-redundant vault storage and must be enabled explicitly. Enable soft delete and immutable vaults so an attacker cannot purge recovery points. For Azure SQL, auto-failover groups give a listener endpoint and a short RPO; geo-restore is the cheapest but has an RPO up to one hour. Cosmos DB uses multi-region accounts with automatic or manual failover.
Worked example: "Retain VM backups for 10 years, restore in the secondary region, survive an attacker deleting the vault." Answer: Recovery Services vault with GRS redundancy, cross-region restore enabled, long-term retention policy, soft delete plus immutability. ASR would not satisfy the retention requirement.
Trap: questions offering Site Recovery as a backup answer, and vice versa.