New See exactly what you're overpaying AWS in under 60 seconds. Try the Calculator for free

RDS Cross-Region Read Replica Costs: Pricing, DR, and Optimization

Learn how RDS cross-region read replicas are priced, what drives replication and data transfer costs, and how to optimize them for disaster recovery and performance.
Updated August 17, 2026
20 min read
RDS Cross-Region Read Replica Costs: Pricing, DR, and Optimization
In this article
Key takeaways
1
Compute and storage create the baseline because the replica is a full RDS DB instance.
2
Cross-Region replication adds transfer charges for seeding and ongoing changes.
3
Db2, MariaDB, MySQL, Oracle, PostgreSQL, and SQL Server support cross-Region replicas, with engine-specific restrictions.
4
Use measured ReplicaLag and business RPO/RTO targets, not assumed lag values.
5
Compare RDS Reserved Instances and Database Savings Plans for steady workloads.
An Amazon RDS cross-Region read replica is a separately billed DB instance in another AWS Region. Total RDS cross-Region read replica cost usually includes destination-Region compute, replica storage, cross-Region data transfer for the initial snapshot and ongoing changes, plus optional backup, IOPS, monitoring, and KMS charges.

AWS does not publish one universal transfer price for every Region pair, so verify the exact route in current RDS pricing.

What does an RDS cross-Region read replica cost?

A read replica is billed at standard RDS DB instance rates in the destination Region. AWS also charges for data transferred out of the source Region when RDS copies the initial snapshot and later database changes. Same-Region read-replica replication does not incur this cross-Region replication transfer charge.
Cost component Billing basis Priority
Replica compute Destination-Region DB instance rate High
Replica storage Provisioned storage plus optional IOPS/throughput High
Cross-Region replication Data sent from source to destination Workload dependent
Backup storage, where supported and enabled Backup configuration and retained data Medium
KMS/monitoring Only when configured Usually low
Note: Backup support varies by engine and version. For example, RDS for SQL Server read replicas do not support automated backups or manual snapshots, and PostgreSQL automated backups on read replicas require version 14.1 or later.
For broader cost modeling, see the RDS pricing calculator and RDS storage cost guide.

Cost formula

AWS charges source-Region data transfer for both the initial snapshot copied when the replica is created and for later database changes, so separate one-time setup cost from recurring monthly cost.
One-time initial snapshot transfer
=
Initial snapshot GB × applicable Region-pair rate
Recurring monthly transfer
=
Monthly changed GB × applicable Region-pair rate
First-month cost
=
Compute + storage + one-time initial snapshot transfer + recurring monthly transfer + backup + optional charges
Steady-state monthly cost
=
Compute + storage + recurring monthly transfer + backup + optional charges

Practical example

Assume 730 instance-hours, 500 GiB of storage, a 500 GB initial snapshot, and 300 GB of replicated changes per month. If your hypothetical inputs are $0.48/hour, $0.115/GiB-month, and $0.02/GB:
Compute
=
730 × $0.48 = $350.40
Storage
=
500 × $0.115 = $57.50
One-time initial snapshot transfer
=
500 × $0.02 = $10.00
Recurring monthly transfer
=
300 × $0.02 = $6.00
First-month cost
=
$350.40 + $57.50 + $10.00 + $6.00 = $423.90
Steady-state monthly cost
=
$350.40 + $57.50 + $6.00 = $413.90 (month two onward)
These are illustrative assumptions, not guaranteed AWS prices. Replace them with current rates for your engine, instance class, storage type, and Regions.

Which RDS engines support cross-Region replicas?

Engine Cross-Region support Can serve reads? Backup support on replica Edition/version condition Multi-AZ replica option Key limitation
Db2 Yes No standby replica (DR-oriented) See AWS docs Version and licensing requirements apply See AWS docs Standby replica does not accept user connections
MariaDB Yes Yes See AWS docs Engine-native asynchronous replication Yes None
MySQL Yes Yes See AWS docs Engine-native asynchronous replication Yes None
Oracle Yes Read-only replicas: yes; mounted replicas: no (DR) See AWS docs Enterprise Edition; CDB architecture constraints Yes Mounted replicas do not accept user connections
PostgreSQL Yes Yes Automated backups require version 14.1 or later Native streaming replication Yes Deleting the source terminates replication; promote or delete manually
SQL Server Yes Yes Not supported (no automated backups or manual snapshots) Enterprise Edition only; specified versions See AWS docs Read replicas cannot take backups
Aurora Separate model N/A N/A N/A N/A Use Aurora-specific cross-Region features or Global Database
AWS currently lists Db2, MariaDB, MySQL, Oracle, PostgreSQL, and SQL Server for cross-Region RDS read replicas. RDS for SQL Server read replicas are currently limited to Enterprise Edition, while PostgreSQL uses native streaming replication.

For deeper PostgreSQL planning, see the RDS PostgreSQL read replica strategy.

When is a cross-Region replica worth the cost?

Disaster recovery

A cross-Region replica provides an asynchronous copy in another Region and can be promoted to a standalone DB instance. AWS says promotion requires a reboot and can take several minutes or longer.

Replication lag can vary significantly, especially across Regions, so a fixed 1–5 second RPO or guaranteed 5–15 minute RTO should not be assumed. Monitor ReplicaLag and test your actual recovery procedure.

Lower-latency reads

A destination-Region replica that accepts read connections can serve read traffic closer to geographically distributed users. Note that RDS for Db2 standby replicas and RDS for Oracle mounted replicas do not accept user connections, so AWS positions those modes for cross-Region disaster recovery rather than read scaling.

Compare a readable replica’s full compute and storage cost with alternatives such as caching before adding another full database instance.

Regional migration

RDS can seed the replica from a snapshot, continue asynchronous replication, and let you promote the replica when you are ready to cut over.
Decision map comparing disaster recovery, lower-latency reads, and regional migration as reasons to use an RDS cross-Region read replica.

How do DR alternatives compare?

Option Ongoing cost Recovery profile Best fit
Cross-Region read replica Highest Low RPO potential, promotion required Critical warm DR target
Cross-Region automated backups Lower Restore required Longer recovery acceptable
Cross-Region snapshot copies Lower Restore from last copy Periodic recovery points
Same-Region Multi-AZ Different purpose Automatic AZ-level failover In-Region high availability
Multi-AZ and cross-Region replication solve different problems. Multi-AZ improves availability within a Region, while cross-Region replicas and replicated backups provide resources in another Region.

How do you optimize cross-Region replica cost?

Follow this sequence:
  1. Confirm the replica is still required for DR, read scaling, or migration.
  2. Right-size compute using CPU, memory, connections, I/O, and ReplicaLag.
  3. Review storage type, IOPS, throughput, and allocated capacity.
  4. Measure actual replicated bytes instead of estimating from total database size.
  5. Compare On-Demand, Reserved Instances, and Database Savings Plans.
  6. Recheck commitments after promotions, migrations, or topology changes.
AWS states that RDS Reserved Instances can save up to 69% compared with On-Demand pricing for steady-state workloads. Database Savings Plans offer up to 35% savings for a one-year usage commitment and can apply across eligible engines, instance families, sizes, deployment options, and regions. The two discount models do not stack on the same usage.

When choosing between them, a stable replica that will keep the same engine, instance family, size, deployment option, and Region for one or three years tends to favor a Reserved Instance. Planned migrations or cross-Region changes tend to favor Database Savings Plans, because that commitment applies across eligible engines, instance families, sizes, deployment options, and Regions for a one-year term.

For commitment planning, see the RDS Reserved Instances guide and Database Savings Plans coverage overview.
Practical callout: Optimize replica compute and storage before focusing only on data transfer. A permanently oversized replica can create avoidable spending every hour.

Topology callout:
For MySQL and MariaDB, AWS notes you can reduce inter-Region transfer charges by creating one cross-Region replica and then adding same-Region replicas from it, rather than sending changes from the source to several remote Regions. Compare “one source to three remote replicas” against “one remote seed replica plus two local replicas.”

Cross-Region replica cost checklist

Verify engine, edition, version, and Region support.
Price compute in the destination Region.
Add storage, IOPS, and throughput costs.
Use the actual Region-pair transfer rate.
Monitor ReplicaLag against your RPO target.
Confirm whether the replica needs Multi-AZ protection.
Review backup retention requirements.
Compare RI and Database Savings Plan economics.
Test the promotion procedure regularly.
RDS supports creating read replicas as Multi-AZ DB instances for MySQL, MariaDB, PostgreSQL, and Oracle, so a replica for these engines should not automatically be treated as permanently Single-AZ. Check the engine-specific documentation before pricing a Multi-AZ replica.

What happens when you promote the replica?

Promotion stops replication, reboots the read replica, and converts it into a standalone DB instance. It continues billing in the destination Region, while replication transfer from the former source stops.

AWS recommends enabling backups and completing at least one backup before promotion when backups are part of your recovery plan.

Deletion behavior also differs by engine. For Db2, MariaDB, SQL Server, MySQL, and Oracle cross-Region replicas, deleting the source promotes the replica. For PostgreSQL, replication is terminated and the replica must be manually promoted or deleted.

DR promotion and cutover runbook

Promotion reboots the instance and can take several minutes or longer depending on replica size, and application traffic must move to the promoted instance. A short ordered runbook:
  1. Stop or fence writes to the source when feasible.
  2. Check ReplicaLag and confirm the replica state.
  3. Promote the replica.
  4. Validate data integrity on the promoted instance.
  5. Change application routing to the promoted instance.
  6. Build a replacement replica for continued protection.

How Does Usage.ai Optimize Cross-Region Read Replica Costs?

A cross-region read replica is a full DB instance in another Region and RDS RIs are region-scoped, so the RI covering your primary never touches it. That replica quietly runs at On-Demand until a matching RI exists in its Region, matched on engine, family, and Single-AZ deployment. It’s the single most expensive RI mistake teams make.

We analyze every primary and replica separately, pin down the Region each one runs in, and buy a matched RI where it actually lives 33% to 69% off On-Demand on covered compute (about 33% on 1-year No Upfront; storage and cross-region transfer stay separate). Recommendations refresh every 24 hours, and we purchase on your approval.

Promote, delete, or resize a replica later? Our cashback protection covers the unused portion in real money, and you pay only a share of the savings we realize.

See how much you can save on RDS read replicas with Usage.ai
Disclosure: AWS pricing, supported versions, and Region availability can change. Numerical examples here are estimates based on stated assumptions, not guaranteed savings or current quotes. Verify production decisions against current AWS pricing and your organization’s actual billing data.
Evaluate with your own data
Run a Free Savings Analysis.

Connect in 15 minutes. No contracts, no infrastructure changes. See your savings before committing.

Frequently asked questions

Does a cross-Region replica cost the same as the primary?

Not necessarily. It uses destination-Region pricing and can have a different supported instance class or storage configuration.

Is cross-Region replication always $0.02 per GB?

No. Treat $0.02/GB only as an example where that rate applies. AWS directs customers to current RDS pricing for cross-Region replication charges.

Does SQL Server support cross-Region read replicas?

Yes. Current AWS documentation supports cross-Region read replicas for supported RDS for SQL Server Enterprise Edition versions and Regions.

Are cross-Region read replicas synchronous?

No. RDS read replicas use asynchronous replication. Cross-Region replicas can experience additional lag, and AWS does not specify one universal lag value.

Can a cross-Region replica use Reserved Instance discounts?

Yes, when the running DB instance matches the applicable Reserved Instance requirements, including Region, engine, and other relevant attributes.

Share
Facebook
X
LinkedIn
Reddit
Cut cloud cost with automation
Latest from our blogs