Managed Service for Prometheus stores metric data from Alibaba Cloud services in dedicated Prometheus instances. Each instance is natively integrated with other cloud services and includes default Grafana dashboards and alerting rules, so you can start monitoring immediately after integration. Through Prometheus instances, you can uniformly collect, store, and display metrics.
The data plane supports two storage types with the same service availability and price:
| Storage type | Redundancy |
|---|---|
| Locally redundant storage | Replicates data within a single zone |
| Zone-redundant storage | Replicates data across multiple zones for higher durability |
Storage redundancy is being upgraded region by region. Regional availability aligns with Simple Log Service. For supported regions, see Storage redundancy.
Prometheus instance creation per service
When you connect a cloud service to Managed Service for Prometheus through Integration Center or default integrations, a Prometheus instance is automatically created to store the collected metrics. The instance creation behavior varies by service type:
| Cloud service | Instance creation behavior |
|---|---|
| Container services, including Container Service for Kubernetes (ACK), Container Compute Service (ACS), ACK Serverless, and ACK One | One Prometheus instance per container cluster |
| Elastic Compute Service (ECS) | One Prometheus instance per ECS instance and virtual private cloud (VPC). This instance stores metrics from the target ECS instance and all other ECS instances within the same VPC. |
| Realtime Compute for Apache Flink | One Prometheus instance per Flink workspace |
| Other cloud services | A single default Prometheus instance named env-cloud-{region}-{uid} per region. All cloud service metrics in that region are stored in this shared instance. |
Each Prometheus instance automatically integrates with Managed Service for Grafana Shared Edition. A corresponding data source named prom-${PrometheusInstanceId} is created in Grafana for querying.
Storage infrastructure upgrade
Starting 00:00:00 (UTC+8) on March 12, 2025, Managed Service for Prometheus is gradually rolling out an upgrade that transitions metric data storage to Metricstores within Simple Log Service projects for improved performance and flexibility. This upgrade does not affect historical data and incurs no additional cost.
Naming formats
After the upgrade, Managed Service for Prometheus automatically creates Metricstores in Simple Log Service projects to store metric data.
Projects:
-
workspace-default-cms-{User ID}-{Region ID}
Metricstores:
| Service | Metricstore name format |
|---|---|
| ACK clusters | aliyun-prom-{Cluster ID} |
| Other services | aliyun-prom-{Managed Service for Prometheus instance ID} |