Amazon Managed Service for Prometheus (Preview)

Use the familiar, flexible Prometheus query language (PromQL) to filter, aggregate, and alarm on metrics, and quickly gain performance visibility for large volumes of metrics labels. AMP supports all metric types: gauge, counter, summary, and histogram.

AMP automatically scales as your ingestion and query needs grow, handling millions of unique time series metrics from large container deployments while maintaining consistent query response times. AMP offers multi-AZ replication within an AWS Region.

AMP integrates with AWS Identity and Access Management for authentication and fine-grained permissions for users and groups. AWS PrivateLink provides easy and secure access to services hosted on AWS, keeping your network traffic within the AWS network. AWS Organizations integration allows for policy control, and API calls are logged to AWS CloudTrail.

You can quickly set up AMP to collect and query metrics from AWS container services, including Amazon EKS and Amazon ECS. AMP also includes an API that makes it easy to securely ingest and query metrics from all of your self-managed Kubernetes clusters, on AWS and on-premises.

Using AMP, you can remove the undifferentiated heavy lifting of running open source Prometheus at scale. This helps significantly reduce the operational costs of configuring, upgrading, and scaling standalone Prometheus servers.

AMP integrates with AWS Distro for OpenTelemetry as a collection agent for Prometheus metrics, and with Amazon Managed Service for Grafana to create rich, powerful data visualizations. This complements other AWS observability services such as Amazon CloudWatch and Amazon Elasticsearch Service.

With AMP, you can monitor containers running on Amazon EKS and Amazon ECS, as well as on-premises. Using Amazon Managed Service for Grafana, you can use the monitoring data collected in AMP to create interactive data visualizations for monitoring, alerts, and dashboards across all your container environments, including host- and application-level monitoring.

Use the powerful PromQL language to detect errors and reduce mean time to resolution. Triggered alerts can be interrogated with PromQL to find systemic versus isolated health and performance issues across all of your containers.

AMP makes it easy to collect and send application metrics from multiple AWS and third-party services without having to manually instrument your application. You can choose from 150+ open source exporters in Prometheus’s library of popular application stacks, including Apache Kafka, Redis, Java/JMX, and NGINX.

AMP is well suited for monitoring high-cardinality data such as video streams and network applications. You can also use it with time series data from IoT devices to analyze data based on labels such as location, category, and user.