Managed Service for Prometheus is a fully managed, Prometheus-compatible monitoring service that collects, stores, and analyzes metrics across your entire stack -- from containers and hosts to cloud services and custom applications. You can use the same PromQL queries, exporters, and dashboards you already know, without managing the underlying infrastructure.
Note Prometheus is an open source monitoring and alerting system. Key features include a multi-dimensional data model, a flexible query language (PromQL), and data visualization. For more information, see Prometheus documentation.
Key concepts
A Prometheus instance is the core logical unit in Managed Service for Prometheus. Each instance includes:
Data collection configurations
A time-series database instance
Dashboard monitoring panels
Alert configurations
Create one or more Prometheus instances to organize monitoring by environment, team, or workload.
How it works
Managed Service for Prometheus provides unified metric collection and storage, with multiple consumption options for visualization, alerting, and analysis.
Data collection
Managed Service for Prometheus collects metrics from six source types:
| Source | When to use | How it works |
|---|---|---|
| Client and server applications | Monitor frontend apps with RUM or backend services with APM. | Metrics flow to Managed Service for Prometheus by default. For custom app metrics, instrument with the Prometheus SDK or OpenTelemetry SDK. |
| Containers | Run workloads on Container Service for Kubernetes (ACK), ACK Serverless, or Container Compute Service (ACS). | A built-in Prometheus agent collects core container metrics automatically. Define ServiceMonitor and PodMonitor resources for custom metric targets. For self-managed Kubernetes clusters or clusters on other cloud providers, register them in ACK One for the same collection capabilities. |
| Hosts | Run workloads on virtual machines or Elastic Compute Service (ECS) instances. | A managed Prometheus agent collects CPU, memory, disk, network, and other OS-level metrics, plus process and container metrics. On ECS, collect custom metrics using the node-exporter textfile format. For on-premises data centers or hosts on other cloud providers, deploy open source node-exporter and Prometheus, then forward data via Remote Write. |
| Cloud services | Monitor Alibaba Cloud services. | Integrate with CloudMonitor through the console. Cloud service integration automatically enriches metrics with instance names and tags for multi-dimensional aggregation, filtering, and alert routing. |
| Custom metrics | Send application-specific metrics from your code. | Use the Prometheus SDK or OpenTelemetry Metrics SDK to instrument your code. Managed Service for Prometheus is compatible with both protocols. |
| Self-managed Prometheus | Centralize data from existing Prometheus instances. | Forward metrics from your existing Prometheus instances via the standard Remote Write protocol. |
Data storage
Managed Service for Prometheus uses tiered storage to balance cost and retention:
| Tier | Description | Billing |
|---|---|---|
| Standard storage | Supports multiple retention periods for active metrics. | Based on data write volume or data reporting volume. |
| Archive storage | When standard storage data expires, it automatically moves to archive storage for long-term retention at lower cost. | Based on storage volume. |
Data consumption
Collected metrics support multiple consumption patterns:
| Capability | Description |
|---|---|
| Visualization | Use prebuilt dashboards for cloud services and common open source components. Import them into Managed Service for Grafana, build custom Grafana dashboards, or connect to DataV via the Prometheus HTTP API. |
| Global view | Query across multiple Prometheus instances and accounts with aggregated views for unified multi-account monitoring. |
| Data processing | Define Recording Rules (compatible with open source Prometheus specifications) for down-sampling and dimensionality reduction. This reduces storage costs and speeds up queries. |
| Alerting | Start with built-in alerting rules for cloud services and common open source components, customize rules with PromQL, or import existing rules from open source Prometheus. Alert events flow into Alert Management for notifications, assignments, and escalations. |
| CloudLens | Unified observability across logs, metrics, and events for core monitoring use cases. |
| Metric analysis | Run ad-hoc queries with full PromQL compatibility. Use metric management tools to inspect metric distribution and identify high-cardinality series. |
| Data export | Stream metrics in real time to Kafka, MaxCompute, or self-managed Prometheus services. |
Benefits
| Benefit | Details |
|---|---|
| Simplified operations | A fully managed monitoring stack with no Prometheus servers to deploy, patch, or scale. Prebuilt dashboards and alerting rules let you start monitoring in minutes. |
| Cost efficiency | Tiered storage and pay-as-you-go pricing based on data volume. Recording Rules for down-sampling help reduce costs compared to self-managed Prometheus deployments. |
| Deep Alibaba Cloud integration | Native support for ACK, ACS, ECS, CloudMonitor, and other Alibaba Cloud services with automatic metric enrichment and prebuilt dashboards. |
| Open standards compatibility | Full PromQL support, Remote Write ingestion, and OpenTelemetry compatibility. Migrate existing Prometheus setups incrementally. |
| Professional support | Technical support backed by a service-level agreement (SLA). |