> ## Documentation Index
> Fetch the complete documentation index at: https://help.nops.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Switch Steady DynamoDB Tables to Provisioned Capacity

> Switch DynamoDB tables with steady traffic from on-demand to provisioned capacity.

On-demand DynamoDB bills each read and write request, which suits unpredictable traffic. A table that serves requests almost every hour at a similar rate costs less with provisioned capacity. nOps finds on-demand tables with steady traffic and estimates the cost of flat provisioned capacity sized for their busiest hours.

| | |
| - | - |
| **Cloud** | AWS |
| **Category** | Database |
| **Service** | Amazon DynamoDB |
| **Action** | Modify |
| **Effort** | Low |
| **Restart needed** | No |
| **Can be rolled back** | Yes |
| **Analysis window** | Last 30 days |
| **Source** | nOps |

## Why it matters

On-demand is often left in place after launch, when nobody yet knew the traffic. For steady tables, provisioned capacity usually costs much less.

## How nOps finds it

nOps reads the last 30 complete days of hourly on-demand request billing for each table. A table is flagged when it served requests in at least 90% of the 720 hours and has at least 30 days of history. nOps sizes flat read and write capacity for the table's 99th-percentile hour at 50% utilization, priced at the public capacity rate scaled by the discount on your own on-demand bill. The saving is the on-demand cost of the last 7 days, scaled to a month, minus that provisioned cost. Tables mostly on reserved capacity, tables mostly paid with credits and savings under \$5 a month are left out.

## Before you start

Permission to update the table and set up auto scaling (`dynamodb:UpdateTable`, Application Auto Scaling).

## How to implement

1. Check the table's traffic in CloudWatch at minute level for spikes within the hour.
2. Switch the table to provisioned capacity with at least the read and write units nOps shows (DynamoDB console: **Tables → your table → Additional settings → Read/write capacity → Edit**, or `aws dynamodb update-table --table-name TABLE --billing-mode PROVISIONED --provisioned-throughput ReadCapacityUnits=RCU,WriteCapacityUnits=WCU`). Set capacity for global secondary indexes too.
3. Turn on auto scaling so capacity follows traffic.
4. Watch for throttled requests for a few days.
5. Update Terraform or CloudFormation so a later deployment does not switch it back.

## Risks and considerations

* Requests above provisioned capacity are throttled. Keep on-demand if traffic within an hour spikes above twice its hourly average.
* AWS limits how often a table can switch between billing modes.
* To roll back, switch the table back to on-demand.

## FAQs

> [accordion] Why 50% utilization?
>
> It leaves room for traffic to double within the hour without throttling. Auto scaling can lower capacity further and save more.

## Related

* [Buy DynamoDB Reserved Capacity](https://help.nops.io/inform/supported-recommendations/buy-dynamodb-reserved-capacity)
* [Switch DynamoDB Table to Standard-IA](https://help.nops.io/inform/supported-recommendations/switch-dynamodb-table-to-standard-ia)
* [Supported recommendations](https://help.nops.io/inform/supported-recommendations.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.