Why did AWS increase the DynamoDB index provisioned read/write capacities significantly beyond the threshold, especially considering that the utilization was lower than expected?

Question

I have a DynamoDB table with the following index read/write capacities:

Read Capacity:
Range: 500 - 5,000
Auto-scaling at 70%

Write Capacity:
Current provisioned units: 500
Range: 100 - 5,000
Auto-scaling at 70%
Current provisioned units: 100

I've experienced scenarios where the utilization is below 70%, yet the capacity is increased to 5,000 and sometimes even exceeds 5,000. What is the reason for this occurrence?

Answer

The reason for this is due to the granularity of CloudWatch metrics. It may seem like you have not exceeded the target utilization, as CloudWatch metrics are averaged across a 1 min period, giving the illusion that you are consuming less capacity then you think.

This article explains the situation in a little more detail.

[This](https://repost.aws/knowledge-center/dynamodb-cloudwatch-metrics-differences) article explains the situation in a little more detail.

Why did AWS increase the DynamoDB index provisioned read/write capacities significantly beyond the threshold, especially considering that the utilization was lower than expected?

Relevant content