Detailed monitoring with CloudWatch in AWS Clean Rooms - AWS Clean Rooms
Types of metricsMetric dimensionsWho can access metricsPricing

Detailed monitoring with CloudWatch in AWS Clean Rooms

Detailed monitoring is an opt-in feature in AWS Clean Rooms. When you create a collaboration and turn on Detailed monitoring, members who can run queries and configured payors can choose to receive detailed observability metrics in their CloudWatch account.

With detailed monitoring, members can monitor query performance, track resource utilization, and create dashboards for operational insights. These metrics help with capacity planning, cost analysis, and troubleshooting performance issues.

Note

Metric availability varies based on configured table data sources:

  • For analyses that reference configured tables with Athena as the data source, the BytesRead and QueryFileScanned metrics always return 0.

  • For analyses that reference configured tables with Snowflake as the data source, the BytesRead, QueryFileScanned, and RecordsRead metrics always return 0.

  • For analyses that reference multiple configured tables with different data sources (e.g., Snowflake and S3-backed AWS Glue tables), BytesRead and QueryFileScanned return metrics only for S3-backed AWS Glue tables. RecordsRead returns metrics for S3-backed AWS Glue tables and Athena tables.

Types of metrics

AWS Clean Rooms publishes two categories of metrics to CloudWatch:

Aggregated detailed monitoring metrics

These metrics provide aggregate-level visibility across analyses:

AnalysisRuntime

Shows the total duration of an analysis in milliseconds.

CRPUConsumed

The CRPU (Clean Room Processing Units) consumed for a particular analysis. For more information about CRPU calculation, see AWS Clean Rooms Pricing.

ConcurrentVCPUs

Concurrent vCPU utilized at a given point in time (related to service quotas).

AnalysesCount

Number of analyses run over a period of time for a specified dimension.

ConcurrentQueries

Concurrent query count at a given point in time (related to service quotas).

Query-level detailed monitoring metrics

These metrics provide detailed insights into individual query execution:

BytesRead

Total data (in bytes) read from source tables for an analysis.

BytesWritten

Total data (in bytes) written for an analysis.

RecordsRead

Total number of records read from source tables for an analysis.

RecordsWritten

Total number of records returned in the analysis result.

QueryFileScanned

Total number of data files scanned for an analysis.

Memory utilization

Memory resource used for an analysis.

vCPU utilization

vCPU resource used for an analysis.

Disk utilization

Storage resource used for an analysis.

Metric dimensions

Metrics are organized by the following dimensions:

  • Analysis Status: Completed

  • Analysis Type: SparkSQL

  • Membership: The membership ID

  • Query: Individual query ID (for query-level metrics)

  • Account/Region: Your AWS account and Region

All metrics are published to the AWS/CleanRooms namespace in CloudWatch.

Who can access metrics

Detailed monitoring metrics are available to:

  • Query runners: Members who can run queries in the collaboration

  • Configured payors: Members designated to pay for query execution costs

Members who only contribute data cannot access detailed monitoring metrics.

Pricing

AWS Clean Rooms does not charge for publishing metrics to CloudWatch. However, standard CloudWatch charges apply for metric storage, dashboard usage, and alarms. For more information, see Amazon CloudWatch Pricing.