Using CloudWatch metrics to investigate throttling issues - Amazon DynamoDB

Using CloudWatch metrics to investigate throttling issues

Below are some DynamoDB metrics to monitor during throttling events. Use these to help locate which operations are creating throttled requests and identify root issues.

  • ThrottledRequests

    • One throttled request can contain multiple throttled events, so events can be more relevant to example compared to requests. For example, when you update an item in a table with GSIs, there are multiple events: a write operation to the table and a write operation to each index. Even if one or more of these events are throttled, there will only be one ThrottledRequest.

  • ReadThrottleEvents

    • Watch for requests that exceed the provisioned RCU for a table or GSI.

  • WriteThrottleEvents

    • Watch for requests that exceed the provisioned WCU for a table or GSI.

  • OnlineIndexConsumedWriteCapacity

    • Pay attention to the number of WCU consumed when adding a new GSI to a table. Note that ConsumedWriteCapacityUnits for a GSI does not include the WCU consumed during index creation.

    • If you've set the WCU for a GSI too low, then incoming write activity during the backfill phase might be throttled.

  • Provisioned Read/Write

    • View how many provisioned read or write capacity units were consumed over the specified time period, for a table or a specified global secondary index.

    • Note that the TableName dimension returns ProvisionedReadCapacityUnits for the table only by default. To view the number of provisioned read or write capacity units for a global secondary index, you must specify both TableName and GlobalSecondaryIndexName.

  • Consumed Read/Write

    • View how many read or write capacity units were consumed over the specified time period.

For more information on DynamoDB CloudWatch metrics, see DynamoDB Metrics and dimensions.