Azure Backup and Azure Site Recovery are sold as two separate services with overlapping but distinct pricing models — and a structural pattern where the storage tier dominates the bill long after the protected instance pricing is sized correctly. The azure backup licensing question that determines whether your backup posture is cost-efficient is the storage tier mix: Locally Redundant Storage (LRS) for most workloads, Geo-Redundant Storage (GRS) only where regulatory or business continuity policy mandates it, and Zone-Redundant Storage (ZRS) as the rarely-best middle option. The second question is retention — backup policies that retain dailies for 365 days and monthlies for 10 years are the most common cost trap, multiplying storage spend without a corresponding business need.
Azure Backup vs Azure Site Recovery
The two services solve different problems and bill differently.
Azure Backup handles backup-and-restore for VMs, SQL Server in VMs, Azure SQL Database, Azure Files, Blob storage, Postgres, and SAP HANA. Pricing has two components: protected instance pricing (a flat per-protected-resource fee that varies by resource size band) plus backup storage (per-GB-month at the chosen redundancy tier). The protected instance fee is fixed; the storage fee scales with retained data volume.
Azure Site Recovery (ASR) handles disaster recovery — continuous replication of running workloads to a secondary region, enabling failover in minutes. Pricing is per-protected-instance per-month plus storage for the replica plus egress during replication.
Most enterprises run both. Backup gives you point-in-time recovery from corruption, ransomware, or human error. Site Recovery gives you regional resilience against an Azure region outage or full datacentre loss. The two are complementary and the costs are independent. A common mistake is treating "we have Backup" as DR coverage — restoring from backup takes hours to days; ASR failover takes minutes.
The storage tier mix that dominates the bill
Backup storage redundancy options:
- LRS (Locally Redundant Storage): three copies in one datacentre. Cheapest. Suitable for most workloads where regional failure is handled separately by ASR.
- ZRS (Zone-Redundant Storage): three copies across availability zones in one region. ~25% more than LRS. Suitable for workloads where you want zone resilience inside backup but full ASR is overkill.
- GRS (Geo-Redundant Storage): replicated to a paired secondary region. ~2x LRS. Suitable for regulatory mandates and high-criticality workloads where you want regional backup resilience without standing up ASR.
The structural mistake we see in 80% of tenants: GRS used as the default for everything. Microsoft's default deployment scripts set GRS; nobody changes it; storage cost compounds at 2x what it could be. For most workloads paired with ASR, LRS backup is the correct choice and GRS is wasted spend.
The Azure portal default for backup redundancy is GRS. The Cost Management recommendations engine never flags this as a downgrade opportunity. The downgrade is one toggle and saves roughly 50% of backup storage spend on the affected vault. The trap is structural — defaults bias toward the more expensive option, and Microsoft has no commercial incentive to surface the cheaper one.
Retention policy decisions that compound
Backup storage cost is the integral of retention. A workload that backs up 500GB daily with 30-day retention stores roughly 15TB; with 365-day retention, 180TB; with 10-year monthly retention added, another 60TB. Each retention tier multiplies the storage line item.
The right retention policy is governance-driven, not "set it and forget it":
- Daily backups: 30 days for most workloads. Some regulatory frameworks demand longer; few demand more than 90 days.
- Weekly backups: 12–13 weeks if quarterly recovery is operationally required.
- Monthly backups: 12–24 months for most workloads. Anything longer should be archived out of Backup to cool or archive tier Blob storage, not retained as live Backup state.
- Yearly backups: only where compliance demands. Most "7-year retention" requirements are satisfied by archived data, not by active backup policy.
| Policy choice | Typical bill impact | Right answer |
|---|---|---|
| GRS for everything | ~2x LRS storage cost | LRS unless regulatory or no ASR |
| 365-day daily retention | ~12x 30-day retention | 30 days; archive longer-term |
| 10-year monthly retention in Backup | 5–15% of total backup spend | Archive to Blob, not Backup |
| Backup VMs without snapshot consistency | Cheaper but unrecoverable | Application-consistent backups |
| ASR replication on every VM | Doubles or triples DR cost | ASR only on production tier |
Reserved capacity for Backup storage
Azure Backup supports reserved capacity for backup storage at three- or five-year commitments. Discount: 18–28% off pay-as-you-go on the storage component. The reservation applies automatically across all vaults in the subscription. For tenants with steady-state backup storage above ~50TB, reserved capacity is a 5-minute purchase decision that captures structural savings.
Reservation sizing follows the same conservative principle as compute reservations: commit to the floor of expected steady-state storage, take pay-as-you-go on the variable top. Backup storage tends to grow predictably as data volume grows, so the floor is easier to identify than for compute.
ASR economics and where it pays back
Azure Site Recovery is expensive per protected instance, but the cost is justified when RTO and RPO matter. The economic question is which workloads get ASR coverage. The right answer is rarely "everything that runs in production". Tier the workloads:
- Tier 0 (revenue-generating, customer-facing, regulatory): ASR with continuous replication, target RTO <30 minutes.
- Tier 1 (internal business-critical): ASR with continuous replication, target RTO <4 hours.
- Tier 2 (important but interruptible): Backup-based recovery, target RTO 24–48 hours. No ASR.
- Tier 3 (dev/test, low-priority): Backup only with short retention. No ASR.
For most enterprises, Tier 0 + Tier 1 accounts for 20–35% of total VM count. Putting ASR on the other 65–80% is structural waste.
Anonymised case study: 6,800-VM manufacturing client
A 6,800-VM industrial manufacturing client ran Azure Backup with GRS storage and ASR on every production VM. Annual cost: $3.9M (Backup) + $4.7M (ASR) = $8.6M. We audited: 92% of backup storage was on GRS where LRS was operationally equivalent (the manufacturer had ASR on the same VMs, so geo-redundancy in backup was redundant). 4,100 VMs were under ASR coverage where the business-tier analysis said only 1,650 needed it. Retention policy was 365 days daily for everything where 30 days plus monthly archive to cool blob was equivalent. We re-engineered: LRS for backup, ASR scope reduced to Tier 0/1 workloads, retention split between active Backup and Blob archive, three-year reserved capacity on the remaining backup storage. New annual cost: $3.4M total — $5.2M saved. RPO and RTO improved on the Tier 0 workloads because the dollars freed up went into more frequent replication.
Where to take this from here
If your backup and DR posture is "GRS for everything, 365-day retention, ASR on every production VM", you are paying 1.8–2.5x what the posture should cost — with no incremental business benefit above the right-tiered configuration. The remediation is sequenced: redundancy tier first, retention policy second, ASR scope third, reserved capacity fourth. For the broader Azure cost picture, the complete Azure cost optimisation guide places backup and DR alongside the rest of the stack. For storage-tier decisions across all Azure workloads, the Azure storage cost optimisation guide is the companion read. For renewal leverage, the EA tier collapse playbook covers how a cost-optimised backup posture reduces MACC burn and changes your negotiation position.