Amazon Managed Service for Prometheus

What is Amazon Managed Service for Prometheus?

A managed monitoring service for container environments.
You can monitor and alert on the performance of containerized workloads using the open-source PromQL without having to scale or manage the underlying infrastructure.
Automatically scale the ingestion, storage, alerting, and querying of operational metrics as workload increases.
Integrated with Amazon EKS, Amazon ECS, and AWS Distro for OpenTelemetry.

With Prometheus query language (PromQL), you can easily obtain performance visibility using a filter, aggregate, and alarm on metrics.
Supports multi-AZ replication within an AWS Region.
Collect and query metrics from container clusters on AWS and on-premises.

Workspace
- A logical space dedicated to storage and querying of metrics.
- Supports one or more workspaces in each Region.
- Metrics ingested into a workspace can only be stored for 150 days.
- Supports multiple rule files.
Rules
- To use rules in Prometheus, you’ll need to create YAML rule files.
- Each rules file is contained under its own namespace.
- Rules are organized into rules groups within a rules file and are evaluated in order from top to bottom.
Recording rules
- You can precompute frequently used or computationally expensive expressions and save the results as a new collection of time series.
- Querying precomputed result is much faster than running the original expression.
Alerting rules
- Create alert conditions based on PromQL and a threshold.
- A notification is sent to the alert manager when a rule triggers the threshold and forwards it to Amazon SNS.
Alert manager
- Supports deduplication, grouping, and routing of alerts to downstream receivers.
- You can also silence and inhibit alerts.
- Configuration file
  - Grouping – collects similar alerts into a single notification.
  - Inhibition – suppresses notifications if certain alerts are already firing.
  - Silences – mute alerts for a specified time.
Templating is stored in the same alert manager configuration file as your alert manager configuration.
Supports tagging for workspaces and rule groups namespaces.

You are charged based on metrics ingested, queried, and stored.
Metrics ingested and stored are only charged when you send metrics to Prometheus.
PromQL Query Samples Processed (QSP) is charged on a per-billion basis.
Storage fees are calculated depending on the compressed size of metric samples and metadata.
There are no charges for data transfer in.