AWS DocumentDB cost: instance hours, storage, and I/O add up

Quick answer

DocumentDB charges per instance hour for each node in the cluster, $0.10 per GB-month of storage, and $0.20 per million I/O requests, plus backup storage. Storage and I/O are shared across the cluster and grow with traffic, so on a read-heavy workload the I/O line can rival compute. Right-size instances and reduce I/O with indexing and caching.

DocumentDB uses the same billing shape as Aurora: you pay for the compute nodes you run, the storage the cluster consumes, and the I/O it performs. Compute is the obvious line; the one that surprises teams is I/O, which scales with how hard the workload reads and writes.

The four lines

Dimension	Rate (approximate, us-east-1)
Instance hours	Per node, by instance class (for example db.r6g.large ~$0.28/hr)
Storage	~$0.10 / GB-month
I/O	~$0.20 / million requests
Backup storage	Free up to cluster size, then per GB-month

Instances are the baseline

Each replica you run is a full instance-hour charge, so a three-node cluster of db.r6g.large costs three times one node around the clock. Right-size the instance class to real peak usage, and add replicas for read scaling or availability only when the read load or SLA needs them, the same discipline as rightsizing without hurting performance.

I/O is the variable you control with your schema

Every read and write that touches storage is billable I/O. A query that scans instead of using an index generates far more I/O than it needs, so indexing hot query paths cuts both latency and cost. Caching frequently read documents in an application cache deflects reads from the cluster entirely, which is the caching-to-reduce-cost lever applied to the database.

Where DocumentDB cost creeps

Oversized instances running at low utilization, missing indexes generating excess I/O, and idle non-production clusters left running overnight are the three common leaks. Stop or downsize dev and staging clusters off-hours, and review the I/O line against query patterns before adding a bigger instance, since more compute does not fix an I/O problem caused by a missing index.

FAQ

How is DocumentDB priced?

By instance hours for each node in the cluster, storage at about $0.10 per GB-month, I/O at about $0.20 per million requests, and backup storage beyond the free allowance. Storage and I/O are shared across the cluster and scale with traffic.

Why is my DocumentDB I/O cost so high?

I/O scales with how the workload reads and writes storage. Queries that scan instead of using an index generate excess I/O, and heavy read traffic drives the line up. Add indexes on hot query paths and cache frequently read documents to reduce it.

Do replicas multiply DocumentDB cost?

Yes. Each replica is a full instance-hour charge, so a three-node cluster costs three times one node. Storage and I/O are shared, but compute is per node, so add replicas only when read scaling or availability requires them.

How do I reduce DocumentDB cost?

Right-size instance classes to real peak usage, add indexes to cut I/O, cache hot reads in an application cache, and stop or downsize non-production clusters off-hours. More compute does not fix an I/O problem caused by a missing index.

Is DocumentDB cheaper than self-managed MongoDB?

It depends on scale and operational load. DocumentDB removes the cost of running and operating MongoDB yourself but charges for instances, storage, and I/O. For small workloads self-managed can be cheaper; at scale the managed operational savings often justify it.

Does C3X estimate DocumentDB cost?

C3X prices the aws_docdb_cluster and aws_docdb_cluster_instance resources from your Terraform, so instance and storage cost appear before deploy. I/O depends on query patterns, which you provide as a usage assumption.

What to do next

Price your cluster before you deploy. C3X reads your Terraform and prices DocumentDB and the rest of your stack against a live catalog. Start with the quickstart.