This project helps you have visibility into high cardinality metrics from an Amazon Managed Service for Prometheus workspace.
Amazon Managed Service for Prometheus - Cardinality insights is maintained by AWS Solution Architects. It is not part of an AWS service and support is provided as best-effort by the community.
To post feedback, submit feature ideas, or report bugs, please use the Issues section of this GitHub repo.
If you are interested in contributing, see the Contribution guide.
This project proposes a serverless solution that you deploy and operate on your account. You give read/write permissions to your workspace and:
- Every 30 min (default, adjustable), it queries your workspace for insights
- Writes a new metric
amp_cardinality_insightsback into your workspace - You can then query this metric for insights, through an optimized Grafana dashboard that we provide
-
Scenario 1: ~75k active series workspace, Amazon EventBridge cron running every 30 min, 1410 metric names costs $1.28 per month (SQS $1.24 + Lambda $0.04)
-
Scenario 2: ~30M active series workspace, Amazon EventBridge cron running every 30 min, 2653 metric names costs $3.20 per month (SQS $2.79 + Lambda $0.41)
To deploy make sure to have the following pre-requisites:
- awscli v2
- aws sam
- IAM permissions to deploy/delete: CloudFormation, Lambda, SQS, EventBridge, create IAM Roles
- A prometheus workspace
- Not required to deploy, but you will need a Grafana environment for visualization
./deploy.sh REGION PROMETHEUS_WORKSPACE_ID [CFN stack name (optional)]- Create an Amazon Managed Grafana Workspace
- Setup your Amazon Managed Service for Prometheus data source
- Import the JSON Grafana dashboard
- Click on a metric to have visibility into the cardinality details
- Explore which cardinalities can be reduced
sam delete
# stack name should be amp-cardinality-insights if you didn't specified your own