Observability with Prometheus and Grafana

Amazon Managed Service for Prometheus (AMP)

Amazon Managed Service for Prometheus (AMP) is AWS’s fully managed, horizontally scalable, Prometheus‑compatible datastore built on Cortex. It ingests metrics from EKS, self‑managed Kubernetes, and standard Prometheus exporters while AWS handles scaling, replication across three Availability Zones, durability, and maintenance. AMP integrates with IAM for authentication, ADOT (https://aws.amazon.com/otel/) for collection, and Amazon Managed Grafana (https://aws.amazon.com/grafana/) for visualization.

• Free Tier & Pricing: AWS Free Tier includes 40 million ingested samples, 10 GB stored, and 200 billion query samples per month. Beyond this, ingestion costs are tiered (about $0.09 per million samples at the first tier). Storage is ~$0.03 per GB‑month, and query processing is ~$0.10 per billion samples. Write bandwidth has no charge; query costs apply after free tier. AWS provides a pricing calculator (https://calculator.aws/). Amazon Managed Grafana is billed separately.

• Storage Retention: Default retention is 150 days and can be configured up to 1095 days (3 years) per workspace. AMP stores metrics at full resolution for the entire retention period with no downsampling. Data is replicated across 3 AZs. For retention beyond 3 years, data must be exported (for example, to S3).

• Query Language & Tools: AMP supports full PromQL and exposes Prometheus‑compatible query APIs using SigV4 authentication (https://docs.aws.amazon.com/AmazonS3/latest/API/sig-v4-authenticating-requests.html). Grafana (managed or self‑hosted) can use AMP as a data source. AMP provides managed Alertmanager and Ruler services: users can upload Prometheus alerting and recording rules for continuous evaluation, with alerts routable through AWS SNS (https://aws.amazon.com/sns/). Any Prometheus‑compatible query tool can query AMP.