DynamoDB Warm Throughput: How to Pre-Warm Tables and Indexes

Written by Rafal Wilinski
Published: November 20th, 2026
DynamoDB warm throughput is the read and write capacity a table or global secondary index (GSI) is prepared to serve immediately across its partitions. AWS launched it on November 13, 2024, for both on-demand and provisioned capacity modes. Pre-warming means setting a higher warm throughput ahead of a known traffic jump. Check WarmThroughput with DescribeTable, then use CreateTable or UpdateTable to raise it if your peak will exceed the current values. AWS launch announcement · AWS warm throughput guide
Defaults and partition limits
Pre-warming prepares DynamoDB's internal partition layout for higher aggregate traffic. DynamoDB also raises the warm values automatically as on-demand usage grows or you increase provisioned capacity. Each physical partition can serve up to 3,000 read units/s and 1,000 write units/s. A table prepared for 10,000 write units/s can still throttle well below that rate if most writes hit one partition. A GSI has its own partitions and can throttle base-table writes through back-pressure. AWS throughput strategies · AWS partition scenarios · AWS GSI back-pressure guide
| Capacity mode | Starting throughput without manual pre-warming |
|---|---|
| On-demand | 12,000 read units/s and 4,000 write units/s for a new table. AWS shows the same default values for a new on-demand GSI. AWS warm throughput scenarios · AWS DescribeTable example |
| Provisioned | Use the RCUs and WCUs you specify at creation as the initial baseline for the table and each GSI. For example, AWS shows a provisioned GSI with 50 RCUs and 25 WCUs reporting 50 read and 25 write units/s before a warm increase takes effect. Check the reported values with DescribeTable. AWS provisioned capacity guide · AWS provisioned GSI example |
Size the peak in units per second. For an item up to 4 KB, one strongly consistent read uses one read unit and an eventually consistent read uses half a unit. A standard write up to 1 KB uses one write unit. Larger items and transactional operations can use more. Estimate base-table and GSI demand separately, including the index writes your application generates. The DynamoDB capacity modes guide covers the capacity and billing terminology. AWS read and write operations · AWS GSI capacity guide
Check warm throughput and readiness
In the DynamoDB console, open Tables, select your table, and find Warm throughput on its details page. DescribeTable returns Table.WarmThroughput and Table.GlobalSecondaryIndexes[].WarmThroughput. Each object has ReadUnitsPerSecond, WriteUnitsPerSecond, and Status. This CLI v2 query shows the warm fields alongside the lifecycle status of the table and each index. AWS console and DescribeTable walkthrough
Pre-warming is asynchronous. AWS says the time depends on the requested throughput and the size of the table or index. Build lead time into the rollout and poll DescribeTable until the warm status is ACTIVE and both reported values meet your targets. Start the traffic increase only after the base table and every relevant GSI pass that check. AWS warm throughput guide
| Change | Status to check |
|---|---|
| Pre-warm an existing table or GSI | WarmThroughput.Status moves from UPDATING to ACTIVE. The existing table and index can remain ACTIVE and continue serving traffic during pre-warming. AWS update walkthrough |
| Create a table or GSI with higher warm throughput | Wait for TableStatus or IndexStatus to reach ACTIVE from CREATING, then check that resource's warm status and values. A GSI added to an existing table can also spend time backfilling. AWS create walkthrough · AWS online GSI operations |
How to pre-warm a DynamoDB table and GSI
Pass WarmThroughput with ReadUnitsPerSecond, WriteUnitsPerSecond, or both. Omit Status from requests. Use the top-level field for the base table. Set the GSI's warm throughput on its own definition or update action. AWS WarmThroughput type · CreateTable API · UpdateTable API
| Operation | Field for the GSI |
|---|---|
CreateTable with a new GSI | GlobalSecondaryIndexes[].WarmThroughput |
UpdateTable on an existing GSI | GlobalSecondaryIndexUpdates[].Update.WarmThroughput |
UpdateTable to add a GSI | GlobalSecondaryIndexUpdates[].Create.WarmThroughput |
In the console, choose Customize settings and Increase warm throughput when creating a table. For an existing table, choose Edit in the Warm throughput section, then Increase warm throughput and save. AWS create instructions · AWS update instructions
AWS CLI v2
Create a new on-demand Orders table with a ByCustomer GSI. The base table starts with targets of 24,000 read and 10,000 write units/s. The index targets 18,000 read and 10,000 write units/s. AWS CLI create-table reference
If Orders and ByCustomer already exist at lower warm values, use this update instead. Read the current values first and pass only the dimensions you want to increase. AWS CLI update-table reference · AWS warm throughput update walkthrough
AWS SDK for JavaScript v3
With @aws-sdk/client-dynamodb, send the same update using UpdateTableCommand. This example assumes both resources exist and their warm values are below the requested targets. AWS SDK v3 UpdateTableCommand · AWS API request shape
boto3
To add a pre-warmed ByWarehouse index to an existing on-demand Orders table, use GlobalSecondaryIndexUpdates with Create. The index can include orders that have a string WarehouseId. boto3 update_table reference · AWS sparse index guide
CloudFormation
On AWS::DynamoDB::Table, set Properties.WarmThroughput for the table and GlobalSecondaryIndexes[].WarmThroughput for each GSI you want to pre-warm. CloudFormation table reference · CloudFormation GSI reference
AWS CDK v2
In a TypeScript CDK app, TableV2 and its GSI properties both accept warmThroughput with readUnitsPerSecond and writeUnitsPerSecond. CDK TableV2 reference · CDK GSI properties
Terraform
With the HashiCorp AWS provider, use warm_throughput on aws_dynamodb_table and inside global_secondary_index. Both blocks use read_units_per_second and write_units_per_second. Terraform AWS provider reference
Quotas, on-demand maximums and provisioned capacity
- Service quotas. You can pre-warm up to the applicable regional table or GSI quota. The adjustable defaults are 40,000 read units/s and 40,000 write units/s per table or GSI. Provisioned capacity also has regional account defaults of 80,000 RCUs and 80,000 WCUs across tables and GSIs. Check your account in Service Quotas before submitting a larger target. AWS allows simultaneous pre-warm requests and documents no limit on the number of tables you can pre-warm at once. Warm values can increase only. AWS service quotas · AWS warm throughput guide
- On-demand maximum throughput.
OnDemandThroughput.MaxReadRequestUnitsandMaxWriteRequestUnitsare separate settings on the table or GSI. If you warm a table to 24,000 read units/s and set its maximum to 10,000, plan for throughput throttling above 10,000. AWS applies this maximum on a best-effort basis, so burst capacity can temporarily let requests through. See on-demand versus provisioned scaling for how the two modes handle spikes. AWS maximum throughput guide · AWS warm throughput scenarios - Provisioned capacity. A table warm to 10,000 writes/s with 2,000 WCUs provisioned still has 2,000 WCUs of configured capacity. Raise the provisioned capacity ahead of the event or adjust auto scaling so capacity is already in place when traffic arrives. Apply the same check to each provisioned GSI. AWS warm throughput scenarios · AWS provisioned capacity guide
Pricing and a worked cost example
AWS charges a one-time fee for each manual increase. Multiply the extra read units by the regional RCU rate and the extra write units by the regional WCU rate. The calculation uses the difference from each resource's current warm throughput. AWS charges for table and index increases separately, and it applies the same model to Standard and Standard-Infrequent Access. Viewing the values and automatic increases carry no warm throughput fee. For current-version global tables (2019.11.21), the settings propagate to replicas and AWS charges in each replica Region. DynamoDB pricing · AWS warm throughput guide
Suppose an on-demand table in us-east-1 is already warm to 12,000 read and 4,000 write units/s. Raising it to 24,000 read and 10,000 write units/s adds 12,000 read and 6,000 write units. At the N. Virginia rates of $0.00013 per added read unit and $0.00065 per added write unit, the one-time charge is $5.46 (12,000 × $0.00013 + 6,000 × $0.00065). Add any GSI or replica fees separately. Use the DynamoDB pricing calculator for ongoing traffic and storage costs. DynamoDB pricing · AWS pre-warming rate example
When to pre-warm DynamoDB
- Launches and peak events. Pre-warm when a cutover, flash sale or launch will bring a predictable step increase across many keys. Compare the expected peak with the current warm values and raise only the resources and dimensions that need more capacity. AWS pre-warming use cases
- Migrations. Prepare the destination before switching live traffic or starting a write-heavy backfill via the DynamoDB API. Include index write demand when sizing the target. If you control the producer rate, gradual ramp-up is another option. AWS pre-warming use cases · AWS on-demand scaling guidance
- Gradual growth and hot keys. Skip paid pre-warming when the expected load fits the current warm values or can ramp gradually. AWS says on-demand traffic can throttle if it exceeds twice the previous peak within 30 minutes. If throttling is concentrated on one key or one index key range, fix the hot partitions by redistributing traffic. Use the DynamoDB throttling guide to identify the affected resource and limit. AWS on-demand scaling guidance · AWS warm throughput scenarios
Frequently Asked Questions
What is DynamoDB warm throughput?
How do I pre-warm an existing DynamoDB table?
How long does DynamoDB pre-warming take?
Does DynamoDB warm throughput cost extra?
Can pre-warming fix a hot partition key?
Dynobase is a Professional GUI Client for DynamoDB
Start your 7-day free trial today
Product Features
DynamoDB Tools
- SSH GUI Client
- Chat With Cloud
- MongoDB GUI Client
- AWS Cost Monitoring
- DynamoDB Query Builder
- DynamoDB Code Examples
- DynamoDB Node.js Queries
- DynamoDB JSON Converter
- DynamoDB UUID Generator
- DynamoDB Local Admin GUI
- DynamoDB Pricing Calculator
- DynamoDB Single-Table Design
- Best DynamoDB GUI Clients
- Dynobase vs DynoTable