Cross-region replication cost: what copying data between regions really costs

Quick answer

Cross-region replication costs two things: cross-region data transfer for every byte copied (priced per GB, typically a couple of cents per GB and more for some region pairs) plus duplicated storage in the destination region (you now pay to store the data twice). The transfer charge is usually the surprise, because it applies to the full initial copy and to every ongoing change. To control it, replicate only what needs replicating, compress before transfer where possible, and weigh whether you need full cross-region replication or a lighter backup strategy.

Replicating data across regions buys durability (surviving a regional outage) and lower latency (serving users from a nearby region), but it is not free, and the cost is easy to underestimate. You pay to move every byte across regions and then to store it a second time. For large or frequently-changing datasets, the cross-region transfer alone can become a meaningful line item.

The two costs of replication

Cost	What it covers
Cross-region data transfer	Every byte copied from source to destination region, per GB
Duplicated storage	Storing the replicated copy in the destination region, per GB per month
Request charges	Replication operations (PUT to destination, etc.)

The transfer charge is the one that surprises people. Cross-region data transfer is priced per GB (typically a couple of cents, more for some intercontinental region pairs), and it applies to the full initial replication plus every ongoing change. Then the destination storage doubles your storage bill for the replicated data. This is the cross-region cousin of ordinary data transfer out.

Why the transfer is the surprise

Storage cost is intuitive, you understand you are keeping two copies. The transfer cost is less visible because it scales with change rate, not just data size. A dataset that changes frequently keeps re-copying deltas across regions, and a large initial replication of an existing dataset can produce a one-time transfer bill that catches teams off guard. Model the initial copy and the ongoing change rate separately when estimating replication cost.

Controlling replication cost

Replicate only what needs it: not every bucket or database needs a cross-region copy, so scope replication to the data where durability or latency genuinely justifies it. Compress data before transfer where the source allows. Consider whether you need full continuous replication or whether periodic cross-region backups (less frequent, cheaper) meet your recovery objectives. For object storage, replicate to a cheaper storage class in the destination (you rarely read a replica) to cut duplicated-storage cost. For read-latency use cases, a CDN may be cheaper than replicating the whole dataset.

Replication in context

Cross-region replication is a durability and latency investment, and for critical data it is worth the cost. The mistake is replicating indiscriminately, copying everything across regions when only a subset needs it, and paying the transfer and duplicated-storage cost on data that does not require it. Weigh the cost against the recovery and latency benefit per dataset. Price your replication configuration against the resource catalog so the transfer and duplicated-storage cost is visible before you turn it on, and fold it into your regional cost planning.

FAQ

What does cross-region replication cost?

Two things: cross-region data transfer for every byte copied (priced per GB, typically a couple of cents and more for some intercontinental region pairs) plus duplicated storage in the destination region (you now store the data twice, paying per GB per month for the copy). Replication operations add small request charges. The transfer is usually the surprise, since it applies to the full initial copy and every ongoing change.

Why is cross-region replication transfer so expensive?

Because cross-region data transfer bills per GB and applies both to the full initial replication and to every ongoing change. A large initial copy of an existing dataset produces a one-time transfer bill, and a frequently-changing dataset keeps re-copying deltas across regions. The cost scales with change rate, not just data size, which is why it catches teams off guard.

How do I reduce cross-region replication cost?

Replicate only what genuinely needs durability or low latency rather than everything, compress data before transfer where possible, consider periodic cross-region backups instead of continuous replication if they meet your recovery objectives, replicate object storage to a cheaper storage class in the destination, and use a CDN instead of full replication for read-latency use cases. Scoping replication tightly is the biggest lever.

Do I pay to store replicated data twice?

Yes. Cross-region replication keeps a copy in the destination region, so you pay destination storage on top of source storage, doubling the storage cost for the replicated data. For object storage, you can cut this by replicating to a cheaper storage class in the destination, since a replica is rarely read directly, which lowers the duplicated-storage portion of the bill.

Is cross-region replication worth the cost?

For critical data, yes, it buys durability (surviving a regional outage) and lower latency (serving users from a nearby region). The mistake is replicating indiscriminately, copying everything when only a subset needs it, and paying transfer and duplicated-storage cost on data that does not require it. Weigh the cost against the recovery and latency benefit per dataset rather than replicating by default.

Does C3X estimate replication cost?

C3X prices infrastructure from your Terraform against a live catalog, so replication configuration and destination-region storage are visible before deploy. Exact transfer cost depends on runtime change rate, but pricing the destination storage and surfacing the replication setup upfront helps you decide whether full cross-region replication is justified before you turn it on.

What to do next

See replication and duplicated-storage cost before you turn it on. C3X reads your Terraform and prices your resources against a live catalog. Start with the quickstart.