Cross region replication copies data from a bucket or database in one AWS region to another, automatically and continuously. The cost is the sum of three things: inter region transfer, charged per gigabyte every time an object is replicated, which for S3 runs at a few cents per gigabyte; the storage of the second copy, which doubles the per gigabyte storage cost for everything replicated; and the request charges on the read and write operations the replication generates. The buyer takeaway is that replication is a deliberate resilience and latency purchase, and the discipline is to replicate the data that earns it and leave the rest in a single region or a cheaper recovery pattern.
Here is what each meter does, and how to size replication against the value it buys.
What actually drives the cost?
The transfer meter is the one teams forget. Every gigabyte that crosses a region boundary to reach the replica is charged, and for write heavy datasets that runs continuously, not once. The storage meter is the obvious one: a replicated dataset is stored twice, so the storage line for that data doubles, and if the replica sits in the same storage class as the source you gain no offset. The request meter is the quiet one: replication generates additional GET and PUT operations that carry their own per request charges, which matter for datasets with many small objects. Together they mean the all in cost of replicated data is well above twice the single region storage figure once transfer and requests are added.
How do you size replication deliberately?
Start from the requirement, not the default. Decide which datasets genuinely need a warm second region copy for recovery time or for serving users in another geography, and which only need a recoverable backup that can be restored more slowly and far more cheaply. For data that must be replicated, tier the replica: a destination in a colder storage class cuts the duplicated storage cost substantially where the replica is for resilience rather than live serving. Filter replication rules so you copy only the prefixes or buckets that need it, rather than whole accounts. And for recovery rather than low latency serving, a periodic copy or a backup with cross region durability often meets the requirement at a fraction of continuous replication.
A worked example
A Fortune 500 retailer had enabled cross region replication broadly after an availability scare, copying many buckets that held logs and intermediate data with no real recovery requirement. The duplicated storage plus the continuous inter region transfer had grown into a material monthly line. Narrowing replication rules to the datasets with a genuine recovery time objective, moving resilience copies to a colder destination class, and replacing continuous replication with periodic cross region backups for the rest cut the replication related spend sharply while preserving the recovery posture that actually mattered. The change was a policy and configuration exercise, not an application one. Figures are verified against billing data and anonymised.
| Meter | What it charges | Lever |
|---|---|---|
| Inter region transfer | Per gigabyte replicated | Replicate only what needs it |
| Duplicated storage | Second copy, per gigabyte | Colder destination class |
| Requests | Extra GET and PUT operations | Fewer, larger objects where possible |
Talk it through with us
If replication was switched on broadly after an incident, it is worth a pass to keep the resilience that matters and drop the duplication that does not. We take zero provider commissions and answer only to you, across AWS, Azure, GCP, and OCI. Our guarantee is plain: we reduce your cloud spend or we reimburse our service fee, on either a Fixed Fee scoped up front or a no risk Gainshare share of verified savings. Book a strategy call to scope it for your estate, and follow more analysis in The Cloud Spend Navigator.
Frequently asked questions
How much does cross region replication cost on AWS?
Is cross region replication worth the cost?
How do you reduce cross region replication costs?
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.