Azure Site Recovery Costs: Breakdown and Optimisation

Azure Site Recovery is a pay-as-you-go disaster recovery service. No upfront hardware, no secondary datacentre, no software licences purchased before you know your actual requirements. But "pay-as-you-go" can obscure the fact that four separate cost components accumulate continuously from the moment you enable protection, and two of them scale directly with how much data your protected workloads generate.

Understanding each component before enabling protection means no surprises on the invoice.

The four cost components

Licence fees per protected instance

ASR charges a per-instance licence fee for each protected VM or physical server, billed hourly. The rate varies by protected workload type (Azure VM to Azure, VMware/Hyper-V to Azure, physical server to Azure). A free trial period of 31 days applies to each new protected instance, after which the licence fee begins.

For Azure-to-Azure replication, the licence fee is the primary fixed cost you control by managing how many VMs are protected. Review the current pricing page for your region (rates in the UK are displayed in GBP) before calculating your protection costs for a given workload set.

Cache storage

ASR replication works by sending changes from protected VMs to a cache storage account in the source region, then forwarding to the target region. You must provision a General Purpose v2 storage account as the replication cache.

The cache storage cost depends on the storage account SKU you choose and the volume of data passing through it. For low-change-rate VMs (office workloads, dev environments), cache costs are modest. For high-I/O workloads (databases, transaction processing), cache storage costs can be significant if you are not monitoring and sizing appropriately.

Target storage

The replicated data lands in a Managed Disk in the target recovery region. You pay for the target Managed Disk continuously, regardless of whether a failover ever occurs. The target disk size matches the source disk, and you pay standard Managed Disk rates for the target region.

For a 1 TB Premium SSD protecting a production database VM, you are paying for a 1 TB Premium SSD in the recovery region at full price, from the moment protection is enabled. Multiply this across a large number of protected VMs and it becomes the largest single cost component.

Optimisation: Use Standard SSD or Standard HDD for target storage on VMs where the recovery RTO allows for the lower-tier disk performance. You can change the target disk type from Premium to Standard without disabling protection, for workloads where a slower disk is acceptable in a DR scenario.

Network charges for replication traffic

Initial replication sends a full copy of each VM's disk data to the target region. This is a one-time egress charge from the source region. For a 1 TB disk, you pay egress on that 1 TB.

Ongoing replication sends only changes (delta replication). The cost is ongoing outbound data transfer from the source region at standard inter-region transfer rates. For low-change-rate VMs, this is negligible. For VMs with high write rates (busy databases, log-heavy applications), ongoing replication traffic costs can be material.

Optimisation: Schedule replication windows during off-peak hours using the replication scheduling feature. Some organisations schedule the bulk delta synchronisation to run overnight (20:00-06:00 GMT) to keep replication traffic costs lower by avoiding peak-hour transfer rates where applicable.

Compute costs during failover

Replication itself does not incur compute costs. Failed-over VMs do. When you run a test failover or actual failover, the target VMs start running and you pay standard VM compute rates for the recovery region.

Test failovers should be temporary: run the test, validate recovery, and clean up. A test failover left running for a week at VM rates for ten protected VMs is an avoidable cost that surprises teams who run tests but forget to tear them down.

Comparing ASR to alternatives

For context, ASR competes primarily with Veeam Data Platform and Zerto for DR workloads.

Veeam uses a perpetual licence model (or subscription). There are no ongoing per-VM replication charges: the licence covers the software, and you provide the storage. For organisations with existing on-premise infrastructure and data centres, Veeam can be more economical. The trade-off is owning and operating the backup infrastructure, which has its own cost and operational overhead.

Zerto uses an annual subscription with per-VM pricing. It supports continuous data protection with near-zero RPO (sub-second), which ASR cannot match (ASR's minimum recovery point objective is approximately 5 minutes for VMware workloads, with typical recovery points every 15-30 minutes for Azure-to-Azure). For workloads requiring RPO in seconds rather than minutes, Zerto is purpose-built.

ASR is cost-competitive for workloads with RPO requirements in the 15-minute to 1-hour range where the managed service benefit outweighs the continuous per-instance licence cost.

Cost management in practice

Right-tier your target storage. Identify which protected VMs genuinely need Premium storage in recovery versus which can tolerate Standard SSD or Standard HDD. Change the target disk type in the replication settings without disabling protection.

Clean up test failover resources promptly. Test failovers create VMs in an isolated network. Build cleanup into your DR test procedure: record the test, confirm recovery objectives are met, then immediately clean up. Set Azure Policy or budget alerts scoped to the ASR test network to catch forgotten test resources.

Audit your protected instance list quarterly. VMs that have been decommissioned but not removed from ASR protection continue to incur licence fees. A quarterly review of protected instances against running VMs in Cost Explorer identifies orphaned protection.

Use Azure Cost Management to separate ASR costs. Tag all ASR-related resources (cache storage, target disks, Recovery Services vault) with a consistent dr-layer or asr tag at creation. This allows you to filter Cost Explorer to show ASR spend distinctly from other Azure costs and track it against your DR budget.

Schedule off-peak replication. For VMs where some replication lag is acceptable, configuring replication to batch during off-peak hours reduces the continuous egress cost from ongoing delta replication. Evaluate whether your RPO requirements allow this before enabling.

Where Critical Cloud comes in

Running Azure Site Recovery correctly, sizing the target environment for cost without sacrificing recovery objectives, and testing failover on a schedule that satisfies FCA, DORA, and PCI DSS evidence requirements, is part of how we operate disaster recovery for regulated businesses. We ensure ASR costs are understood, optimised, and tracked within a broader Azure spend model. As the world's first Powered by Datadog accredited partner, we monitor replication health, recovery point lag, and test failover outcomes as live signals alongside the rest of the Azure estate. See how Critical Support works.