.png)
AWS
Microsoft Azure
Google Cloud
A running instance can scale down to 0.5 DCU, but not to zero. Standard also charges $0.10/GB-month for storage and $0.20 per million database I/Os. I/O-Optimized charges $0.30/GB-month for storage and includes database I/O. Confirm current regional rates on the AWS DocumentDB pricing page before approving an estimate.
All rates in this article are US East (N. Virginia) examples. AWS pricing varies by Region, so check the AWS pricing page before using these figures for budgeting.
DocumentDB Serverless pricing at a glance
AWS measures Serverless compute in DocumentDB Capacity Units, or DCUs. One DCU provides approximately 2 GiB of memory with corresponding CPU and networking. Capacity moves in fine-grained increments of 0.5 DCU.| Billing component | Current rate or rule | What it means |
|---|---|---|
| Standard compute | $0.0822/DCU-hour | Billed per second; 10-minute minimum after a billable status change |
| I/O-Optimized compute | $0.0905/DCU-hour | Database I/O included |
| Minimum running capacity | 0.5 DCU | No scale-to-zero while running |
| Capacity granularity | 0.5 DCU increments | Fine-grained scaling |
| Standard storage | $0.10/GB-month | Data, indexes, change stream data |
| Standard database I/O | $0.20 per million I/Os | Separate variable charge |
| I/O-Optimized storage | $0.30/GB-month | No separate database I/O charge |
| Backup storage | Free up to 100% of total cluster storage in the Region | Excess backup is region-priced |
| Inter-AZ replication transfer | No charge | Replication traffic between cluster instances |
For a broader view of provisioned and Serverless billing dimensions, see our DocumentDB pricing breakdown.
How to estimate your monthly bill
Before calculating, confirm four inputs:Region.
Number of Serverless writer and reader instances.
Storage configuration.
Backup and data-transfer assumptions.
Sum the DCU-hours consumed by every Serverless writer and reader.
Multiply those DCU-hours by the applicable Standard or I/O-Optimized compute rate.
Add database storage using the selected configuration's GB-month rate.
For Standard, divide database I/O requests by one million and multiply by $0.20.
Add any billable backup storage and applicable AWS-wide data-transfer charges.
Standard vs I/O-Optimized costs
Standard is designed for low to moderate I/O consumption. I/O-Optimized is designed for I/O-intensive workloads or teams that value more predictable billing. AWS says Standard may be suitable when I/O is below 25% of total cluster spend, while I/O-Optimized can offer better price performance when I/O exceeds 25%.Treat 25% as a screening heuristic, not a guaranteed mathematical crossover. Serverless compute and storage rates differ between the configurations, so exact break-even depends on DCU-hours, stored data, backup usage, and I/O volume.
Standard read I/O is counted when an 8 KB page is read from storage; cached pages do not add another storage read. Write I/O is counted in 4 KB transaction-log units. Scans, bulk operations, change streams, TTL indexes, and garbage collection can increase I/O.
A three-question storage decision
Is the working set mostly memory-resident? If yes, Standard deserves the first comparison. If large scans and repeated cache misses dominate, model I/O-Optimized.
Is write activity heavy and sustained? If yes, include write I/O and garbage-collection I/O in the Standard estimate.
What does the complete monthly model show? Calculate both configurations with the same DCU-hours, storage, backup, and workload assumptions.
AWS examples, extended monthly
AWS’s official Serverless example uses a workload that rises from 0.5 to 7 DCUs, runs for 30 minutes, then takes about three minutes to return to 0.5 DCU. AWS calculates compute costs of $0.32 with Standard and $0.35 with I/O-Optimized.Those are compute-only figures. For FinOps planning, combine compute with storage and I/O.
Illustrative monthly example
Assumptions: 7 DCUs average for 730 hours, 100 GB of storage, 500 million database I/Os, and no billable backup under the regional allowance. This is an example, not a forecast.
| Cost component | Standard | I/O-Optimized |
|---|---|---|
| Compute | 7 × $0.0822 × 730 = $420.04 | 7 × $0.0905 × 730 = $462.46 |
| Storage | 100 × $0.10 = $10.00 | 100 × $0.30 = $30.00 |
| Database I/O | 500 × $0.20 = $100.00 | Included = $0 |
| Backup | $0 | $0 |
| Total | $530.04 | $492.46 |
The 0.5 DCU baseline cost
DocumentDB Serverless can scale an idle writer or reader to 0.5 DCU when MinCapacity is configured to 0.5. At that floor, Standard compute for one continuously running instance is approximately:Eligible development and test clusters can be stopped for up to seven days. Compute charges stop, while applicable storage and backup charges continue. AWS automatically restarts a stopped cluster after seven days.
Serverless vs provisioned cost
Serverless suits variable demand; provisioned capacity can become more attractive when demand stays high and predictable.A compute-only reference illustrates the trade-off. AWS’s DocumentDB pricing example lists db.r5.large at $0.277/hour. Dividing that rate by Standard Serverless compute gives:
Track ServerlessDatabaseCapacity and DCUUtilization in CloudWatch before changing architecture. AWS documents both metrics for understanding actual Serverless capacity consumption. Our AWS Cost Explorer guide can help connect that operational view with broader cost analysis.
Database Savings Plans for Serverless
AWS Database Savings Plans explicitly include Amazon DocumentDB Serverless. AWS says the plans can provide up to 35% savings on eligible serverless database usage in exchange for a consistent hourly usage commitment over a one-year term with no upfront payment. Eligible usage beyond the hourly commitment is billed at On-Demand rates.Size commitments against durable hourly eligible spend, not traffic peaks or one cluster in isolation.
What a Database Savings Plan does not cover
Keep the commitment model separate from the full DocumentDB bill. Model storage, backup, Standard I/O, applicable data transfer, and usage beyond the commitment separately rather than assuming the headline discount applies to the whole bill.Our Database Savings Plans guide explains commitment mechanics in more detail, and our full Database Savings Plans support announcement covers the AWS database services supported by our automation.
How Usage.ai handles DocumentDB Serverless costs
Usage.ai helps teams identify where DocumentDB Serverless costs can be reduced by analysing sustained DCU usage, idle capacity, and commitment opportunities.For steady workloads, we help evaluate whether provisioned capacity may be more cost-effective and identify eligible Database Savings Plans opportunities, which can provide savings on committed usage.
For dev and staging clusters sitting at the 0.5 DCU minimum, stopping idle environments can eliminate compute charges while storage and backup costs continue.
For eligible AWS commitments outside Database Savings Plans, Flex Insured Commitments can provide up to 57% savings associated with three-year AWS commitment coverage while reducing long-term commitment risk. If a covered commitment costs more than equivalent On-Demand usage, Usage.ai provides cashback protection, subject to eligibility and program terms.
Our fee is based on realized savings.
For broader commitment mechanics, see our AWS Savings Plans guide.
Frequently asked questions
How much does DocumentDB Serverless pricing start at?
In US East (N. Virginia), Standard Serverless compute is $0.0822 per DCU-hour and I/O-Optimized compute is $0.0905 per DCU-hour. A continuously running instance configured with a 0.5 DCU minimum has a Standard compute floor of about $30 for a 730-hour month, before storage and I/O.
What is the difference between Standard and I/O-Optimized?
Standard has lower compute and storage rates but charges database I/O separately. I/O-Optimized has higher DCU and storage rates but includes database I/O. AWS uses a 25% I/O-spend guideline, but the better choice comes from comparing complete monthly costs for your workload.
Can I switch storage configurations whenever I want?
You can switch an existing cluster to I/O-Optimized once every 30 days. AWS allows you to switch back to Standard at any time. Check the cluster configuration or describe-db-clusters before scheduling another change to I/O-Optimized.
Does DocumentDB Serverless scale to zero?
No. An idle Serverless writer or reader can scale to 0.5 DCU when MinCapacity is configured to 0.5. AWS notes that this idle state is intended to remain ready to scale, not to provide enough CPU for most production workloads.
Do Database Savings Plans cover DocumentDB Serverless?
Yes. AWS lists Amazon DocumentDB Serverless as eligible. Database Savings Plans can provide up to 35% savings on eligible serverless usage with a one-year consistent-usage commitment and no upfront payment. Usage beyond the hourly commitment remains at On-Demand rates.
