How to Leverage Alibaba Cloud Prometheus for Fine‑Grained Cloud Product Monitoring

Motivation for Enhanced Cloud Metrics

Enterprises using Alibaba Cloud services (ECS, RDS, Redis, etc.) often need fine‑grained metric processing and flexible alerting that the default CloudMonitor cannot provide. Typical requirements include top‑N CPU usage per region, Kafka consumer‑group lag alerts, and disk‑space‑based cleanup predictions.

Why Custom Prometheus Exporters Are Difficult

Building a dedicated exporter for each cloud product requires developers to use one of the four official client libraries (Go, Java, Python, Ruby) to call Alibaba Cloud’s real‑time export API. This adds development effort, introduces extra runtime threads, and consumes additional CPU, slowing feature delivery.

Alibaba Cloud Managed Prometheus Service

Alibaba Cloud offers a fully managed Prometheus service that integrates with over 25 cloud products and provides pre‑built multi‑dimensional Grafana dashboards. The service supports two integration modes:

Enterprise Cloud‑Monitoring Integration : The service pulls metrics from the CloudMonitor API. Storage and ingestion are free; only data‑transfer fees apply.

Cloud‑Product Self‑Monitoring Integration : An exporter pod runs inside the product’s own Kubernetes cluster and pushes metrics directly to the managed Prometheus instance.

Enterprise Cloud‑Monitoring Integration

In this mode, an Exporter pod is deployed in the managed Prometheus K8s cluster. It calls the CloudMonitor real‑time export API to collect metrics. An Arms Prometheus Agent handles data collection, Alarm templates define alert rules, and Grafana provides default dashboards.

Supported products (growing list) include:

Compute: ECS

Storage: SLS, OSS

Network: ALB, API Gateway, Connector, CDN, CEN, DCDN, Cloud NAT, EIP

Big Data: E‑MapReduce, Elasticsearch, Logstash

Databases: PolarDB, RDS (PostgreSQL, MySQL, SQLServer), Redis, Hologres, ADB, DRDS, DTS

Security: WAF

Using the Enterprise Integration

After enabling the integration, users open the Prometheus console, go to the "Access Center", select a cloud product, and the system creates a dedicated Prometheus instance named "Enterprise Cloud Service". The product’s metric page shows basic information and recent 10‑minute metric volumes.

Grafana Dashboards and Alerts

Grafana automatically displays a default dashboard for each integrated product. Users can create new alert rules using built‑in templates or custom PromQL expressions.

Sample Alert Rules (PromQL)

AliyunEcs_cpu_total{} > 80

AliyunEcs_memory_usedutilization{} > 90

AliyunEcs_diskusage_utilization{} > 90

AliyunEcs_diskusage_free{} < 1024*1024*1024*5

Cloud‑Product Self‑Monitoring Integration

For products that already expose their own observability data, this mode consolidates those metrics into a single Prometheus view. The deployment consists of an Exporter pod inside the product’s Kubernetes cluster, the same Arms Prometheus Agent, alert templates, and Grafana dashboards.

Supported self‑monitoring products include ClickHouse, Lindorm, MongoDB, RabbitMQ, Kafka, RocketMQ, EDAS, MSE, SAE, AHAS, Grafana Service, and PTS.

Example: Monitoring MSE Cloud‑Gateway

Create the MSE instance and enable the ARMS Prometheus option.

Open the product’s monitoring entry in the console.

Advanced Features and Best Practices

While the shared Grafana edition provides basic dashboards, teams can upgrade to Grafana Professional or Enterprise editions for custom dashboards, advanced alerting, and higher query limits.

Summary of Alibaba Cloud Managed Prometheus

Alibaba Cloud Prometheus is a fully managed, cloud‑native monitoring service built on the open‑source Prometheus project. It natively integrates common Alibaba Cloud services, supports major open‑source components, and provides out‑of‑the‑box Grafana dashboards, intelligent alerts, and optimized probe performance. By offering a unified metrics layer, it reduces operational overhead, accelerates issue detection, and improves overall reliability for enterprise workloads.