AWS
Microsoft Azure
Google Cloud
Amazon DocumentDB Serverless launched in general availability on July 31, 2025. It applies the auto-scaling model to DocumentDB — AWS’s MongoDB-compatible fully managed document database. Instead of choosing a fixed instance type (db.r5.large, db.r6g.xlarge), you let DocumentDB automatically adjust compute capacity in real time based on actual workload demand.
The billing unit is the DocumentDB Capacity Unit (DCU). You pay for the DCUs your cluster consumes, measured per second. Each serverless writer or reader scales within its configured bounds; every instance’s DCU consumption is billed separately. When load drops, capacity scales back down. You pay only for what is actually consumed, not for a fixed instance running 24 hours a day.
See exactly what you’re overpaying in under 60 seconds. Try the Calculator for free →
DocumentDB Serverless Pricing at a Glance
| Component | Rate | Notes |
| DCU (Standard storage) | $0.0822/DCU-hr | Billed per second. 1 DCU = ~2 GiB memory + CPU + networking. Minimum 0.5 DCU. |
| DCU (I/O-Optimized storage) | $0.0905/DCU-hr | Higher DCU rate, I/O charges included. Better price predictability for I/O-heavy apps. |
| Minimum capacity | 0.5 DCU | Always-on floor. At 0.5 DCU continuously: ~$30/month in compute. |
| Capacity granularity | 0.5 DCU increments | Fine-grained scaling. Smooth cost curve between min and peak. |
| Storage (Standard) | $0.10/GB-month | Includes 3-AZ replication. Pay for one logical copy. Auto-scales — no pre-provisioning. |
| Database I/O (Standard only) | $0.20 per million I/O requests | Read I/O: 8K pages = 1 I/O. Write I/O: 4K transaction log units. Verify rate at aws.amazon.com/documentdb/pricing/. |
| Storage (I/O-Optimized) | $0.30/GB-month | I/O charges included. Choose when I/O cost would exceed 25% of total cluster spend under Standard. |
| Backup storage (free tier) | Free to 100% of DB size | Automated backups free up to total cluster storage. Above free tier: $0.021/GB-month. |
| Backup storage (above free) | $0.021/GB-month | Only charged above the free tier threshold. |
| Inter-AZ data transfer | $0.00 | Data transferred between cluster instances across AZs is free. |
| T3/T4g CPU credit overage | $0.09/vCPU-hr | Applies only to provisioned T3/T4g instances, not DCU-based serverless instances. |
| Database Savings Plans | Eligible | DocumentDB is explicitly listed for Database Savings Plans on AWS pricing page. Up to 35% savings. |
| Free trial | 30 days | Provisioned t3.medium trial: 750 hours, 30 million I/Os, 5 GB storage and 5 GB backup for eligible new customers over 30 days. This is not free serverless DCU usage. |
All rates: US East (N. Virginia), October 15, 2026. Source: AWS official DocumentDB pricing page. Verify at aws.amazon.com/documentdb/pricing/ — rates change.
Source: AWS official DocumentDB pricing page (aws.amazon.com/documentdb/pricing/), October 15, 2026. Minimum cost estimate: 0.5 DCU x $0.0822 x 730 hrs = $30/month compute only, Standard storage. Verify all current rates before estimating.
What Is a DCU?
A DocumentDB Capacity Unit (DCU) is the billing unit for DocumentDB Serverless. Each DCU provides approximately 2 GiB of memory with corresponding CPU and networking resources — similar to the resource allocation in DocumentDB provisioned instances. Source: AWS official DocumentDB pricing page.
The DCU model means you pay for memory-equivalent capacity, not for specific vCPU counts or network bandwidth separately. The cluster is billed per second at the current DCU level. If your cluster runs at 4 DCUs for exactly 15 minutes, the charge is 4 x $0.0822 x (15/60) = $0.0822.
DocumentDB Serverless scales capacity in fine-grained 0.5 DCU increments. The minimum is 0.5 DCU — there is no true scale-to-zero as of October 15, 2026 for Serverless DocumentDB. Unlike Aurora DSQL (which scales to zero when idle), DocumentDB Serverless maintains a minimum of 0.5 DCU at all times. The minimum compute cost at 0.5 DCU running continuously: 0.5 x $0.0822 x 730 = approximately $30/month.
No scale-to-zero: unlike Aurora DSQL, DocumentDB Serverless does not scale to zero. The 0.5 DCU minimum means there is always a baseline compute charge. For applications that are truly idle for long periods, you can manually stop the cluster (up to 7 days) — storage charges continue but compute stops. Restart when needed.
Also read: Amazon Keyspaces Serverless: The Complete 2026 Guide
Standard vs I/O-Optimized: Which Storage Configuration?
DocumentDB Serverless offers two storage configurations, the same choice available for provisioned DocumentDB. The DCU rate differs between them.
Standard Storage ($0.0822/DCU-hr)
Standard is the default. You pay the lower DCU rate but are charged separately for I/O operations. Read I/Os are billed at 8K page granularity — one I/O per 8K page read from storage. Write I/Os are billed in 4K transaction log units. Choose Standard when your application reads its working set from memory most of the time (well-indexed, right-sized) and I/O costs stay below 25% of your total cluster spend.
I/O-Optimized Storage ($0.0905/DCU-hr)
I/O-Optimized costs $0.0905/DCU-hour plus $0.30/GB-month and includes I/O. At the same capacity usage and storage, its extra charge is $0.0083 × monthly DCU-hours + $0.20 × average stored GB. Compare that premium with Standard I/O charges. For 2 average DCUs, 730 hours and 50 GB, the premium is $22.118, so break-even is 110.59 million I/Os. AWS’s 25% guidance is not an exact universal crossover.
Official Worked Examples (Directly from AWS)
Scenario: a workload that needs 7 DCUs, runs for 30 minutes, then scales down from 7 DCUs to 0.5 DCUs over approximately 3 minutes.
Example 1: Standard Storage
| Usage | Calculation | Cost |
| Scale to 7 DCUs instantly | ||
| Run at 7 DCUs for 30 minutes | 7 DCUs x $0.0822/hr x 30/60 hr | $0.29 |
| Scale down from 7 to 0.5 DCUs in 3 minutes | 7 DCUs x $0.0822/hr x 3/60 hr | $0.03 |
| Compute total (Standard) | $0.32 |
Example 2: I/O-Optimized Storage
| Usage | Calculation | Cost |
| Scale to 7 DCUs instantly | ||
| Run at 7 DCUs for 30 minutes | 7 DCUs x $0.0905/hr x 30/60 hr | $0.32 |
| Scale down from 7 to 0.5 DCUs in 3 minutes | 7 DCUs x $0.0905/hr x 3/60 hr | $0.03 |
| Compute total (I/O-Optimized) | $0.35 |
Note: these examples cover compute only — storage and I/O (for Standard) are additional. Storage at $0.10/GB-month and I/O at per-million rates are billed separately for Standard. I/O-Optimized has no separate I/O charge. Source: AWS official DocumentDB pricing page.
DocumentDB Serverless vs Provisioned: When Each Is Cheaper
DocumentDB Serverless is cost-effective for variable workloads where average DCU utilization is significantly below the provisioned equivalent. For steady, predictable workloads, provisioned instances are typically cheaper because the fixed per-instance-hour rate is lower than the equivalent DCU cost at sustained high utilization.
Break-even example: a DocumentDB provisioned db.r5.large (2 vCPU, 16 GB) costs approximately $0.277/hr on-demand in US East. The equivalent DCU capacity for 16 GiB is approximately 8 DCUs (at 2 GiB/DCU). DocumentDB Serverless at sustained 8 DCUs: 8 x $0.0822 = $0.658/hr — 2.4x more expensive than the equivalent provisioned instance. Serverless saves money only when average DCU consumption is well below the provisioned equivalent.
For the r5.large price comparison, compute-only break-even is $0.277 ÷ $0.0822 = about 3.37 average DCUs. That is 42.1% of an 8-DCU memory comparison, not a universal 50% rule. Equal memory does not establish equal workload performance. Include replicas, I/O, storage and actual commitment eligibility. DocumentDB does not offer the provisioned RI discount assumed in the original comparison; eligible Generation 7+ instances and serverless can use Database Savings Plans.
| Scenario | Recommended | Reason | Monthly saving |
| Variable workload, 60-80% lower at night/weekends | Serverless | Scales down during off-peak, pay only for DCUs consumed. | Workload-dependent; no fixed percentage |
| Dev/test with intermittent queries | Serverless | Low DCU floor during idle periods. Stop cluster overnight for zero compute. | Compare measured DCU-hours and any scheduling alternative |
| New application, unknown workload | Serverless | No capacity planning needed. Set wide min/max bounds and observe CloudWatch DCU. | Avoids over-provisioning waste |
| Steady high-throughput production, 24/7 | Provisioned, with DSP only where eligible | Sustained DCU cost exceeds provisioned rate. Fixed instance cheaper at high utilization. | Use an actual eligible quote and benchmarked workload |
| Flash sales, gaming events, periodic promotions | Serverless | Scales instantly to handle peak without pre-provisioning. No idle waste between events. | Eliminates cost of provisioning for peak year-round |
Source: analysis from verified rates and AWS official DocumentDB Serverless use case documentation (docs.aws.amazon.com/documentdb/latest/developerguide/docdb-serverless-how-it-works.html). Verify rates at aws.amazon.com/documentdb/pricing/.
Also read: AWS Database Savings Plans
DocumentDB Version 3.6 Extended Support: The 2026 Cost Trap
Version 3.6 Extended Support is relevant to older provisioned clusters, not DocumentDB Serverless. Serverless supports eligible 5.0 and 8.0 engine versions and cannot run 3.6 or 4.0. The earlier claim of a 3.6 Extended Support surcharge per serverless DCU-hour was incorrect. Source: AWS serverless requirements.
For a separate provisioned r5.large example in N. Virginia, the version 3.6 surcharge effective July 1, 2026 is $0.1108 per vCPU-hour for years 1–2, then $0.2216 for year 3. Two vCPUs add $0.2216/hr initially, taking the $0.277 base rate to $0.4986/hr. Upgrade affected provisioned clusters after compatibility testing to stop future surcharges.
How DocumentDB Serverless Scaling Works
DocumentDB Serverless monitors your workload continuously and adjusts capacity in real time. The scaling mechanism is designed to have minimal overhead so it responds quickly to changes in query load. Source: AWS official DocumentDB Serverless documentation.
Scale-up happens instantly — the cluster can scale to handle hundreds of thousands of transactions per second without manual intervention or service interruption. Scale-down is gradual: returning to minimum capacity after peak load takes approximately 3 minutes, as shown in the AWS pricing examples. This prevents oscillation from brief traffic spikes while ensuring efficient cost tracking after sustained load drops.
The minimum capacity of 0.5 DCU is always maintained. This ensures the cluster can accept connections and respond to queries immediately without any resume latency — unlike true serverless products that require cold-start time after full scale-to-zero. The trade-off is the minimum $30/month baseline compute cost. For applications that go days or weeks between queries, manually stopping the cluster is the way to eliminate this baseline cost.
MongoDB Compatibility and Supported Features
DocumentDB Serverless is MongoDB API-compatible and supports all features of DocumentDB provisioned clusters. This means you can use existing MongoDB drivers, tools, and application code without changes. Source: AWS official announcement (July 31, 2025).
Supported features include Multi-AZ, replicas, monitoring and I/O-Optimized storage, subject to engine, patch and Region support. Serverless currently requires supported 5.0 or 8.0 engine versions; 3.6 and 4.0 are not serverless options.
DocumentDB is not a fully compatible MongoDB replacement — it supports a MongoDB-compatible API rather than the full MongoDB feature set. Applications using advanced MongoDB features (full MongoDB aggregation pipeline operators, certain index types) should verify compatibility before migrating. Use the AWS Schema Conversion Tool or test your specific queries against DocumentDB before committing.
See exactly what you’re overpaying in under 60 seconds. Try the Calculator for free →
How Usage.ai Handles DocumentDB Serverless Costs
Usage.ai monitors DocumentDB Serverless DCU consumption from CloudWatch and identifies two primary optimization opportunities: clusters running at sustained high DCU utilization where provisioned instances would be cheaper, and clusters eligible for Database Savings Plans commitment.
Database Savings Plans cover DocumentDB and provide up to 35% savings on a 1-year commitment. For teams running DocumentDB Serverless at consistent minimum spend patterns, the DSP commitment delivers savings on the stable floor of DCU spend without requiring a specific instance type commitment.
For teams with separate provisioned DocumentDB 3.6 clusters, include Extended Support billing already effective from July 1, 2026 in the upgrade analysis. Do not apply that surcharge to serverless DCUs.
If any DSP commitment for DocumentDB becomes underutilized — because a cluster is decommissioned or migrated — Usage.ai provides cashback on the unused commitment in real money. Fee: percentage of realized savings only.
Also read: See how Usage.ai optimizes DocumentDB Serverless and Database Savings Plans
Frequently Asked Questions
1. What is DocumentDB Serverless?
Amazon DocumentDB Serverless is an auto-scaling configuration for DocumentDB (MongoDB API-compatible) that launched in general availability on July 31, 2025. It automatically scales compute in DocumentDB Capacity Units (DCUs) based on workload demand, with no manual instance sizing required. Billing is per second at $0.0822/DCU-hour (Standard) or $0.0905/DCU-hour (I/O-Optimized). It supports all DocumentDB features including Multi-AZ, read replicas, and Performance Insights. Source: AWS official announcement and pricing page.
2. How much does DocumentDB Serverless cost?
$0.0822/DCU-hour for Standard storage or $0.0905/DCU-hour for I/O-Optimized, billed per second. Minimum 0.5 DCU per serverless instance: $30.003, approximately $30/month, over 730 hours (Standard). Storage: $0.10/GB-month (Standard) plus I/O charges, or higher rate with I/O included (I/O-Optimized). Backup storage: free up to 100% of cluster size, $0.021/GB-month above. Database Savings Plans eligible for up to 35% savings. Source: AWS official pricing page, October 15, 2026.
3. Does DocumentDB Serverless scale to zero?
No. The minimum capacity is 0.5 DCU, which is always maintained. DocumentDB Serverless does not scale to zero — there is always a baseline compute charge of approximately $30/month (Standard) for continuous operation. To temporarily eliminate compute charges, you can manually stop the cluster for up to 7 days, after which AWS automatically restarts it. Storage charges continue while the cluster is stopped. Source: AWS official DocumentDB pricing page.
4. What is a DCU?
A DocumentDB Capacity Unit provides approximately 2 GiB of memory with corresponding CPU and networking resources. Capacity scales in 0.5 DCU increments, and billing is per second at the DCU count currently allocated. 4 DCUs provides approximately 8 GiB of memory. A db.r5.large has 16 GiB, and equal memory alone does not prove equal CPU or application performance. Source: AWS official DocumentDB pricing page.
5. When is DocumentDB Serverless cheaper than provisioned?
When average DCU consumption is significantly below the provisioned equivalent. A variable workload averaging 2 DCUs on a cluster that peaks at 8 DCUs costs approximately $120/month on Serverless (Standard) versus $202.21/month for one continuously running provisioned db.r5.large. For steady workloads averaging 8 DCUs continuously, Serverless at $0.658/hr exceeds a provisioned db.r5.large at $0.277/hr by 2.4x. Use CloudWatch DCU metrics after 30 days to determine your actual average utilization.
6. What changed with DocumentDB 3.6 in 2026?
DocumentDB 3.6 Extended Support applies to provisioned instances; Serverless supports eligible 5.0 and 8.0 versions and cannot run 3.6. In N. Virginia, affected r5 instances add $0.1108/vCPU-hour in years 1–2 and $0.2216 in year 3. Billing began July 1, 2026. There is no 3.6 serverless DCU surcharge to include.
