Managed Service for Prometheus collects metrics from Alibaba Cloud services through two channels. Choose the channel that fits your monitoring requirements:
| Cloud Monitor metrics | Advanced metrics | |
|---|---|---|
| Collection method | Relayed from Cloud Monitor | Scraped directly by Managed Service for Prometheus |
| Service coverage | All cloud services | Specific services only (see the Supported services and provisioning policies section below) |
| Regional scope | Cross-region (all regions connected automatically) | Depends on provisioning policy (same-region, by cluster, or by workspace) |
| Collection interval | Standard Cloud Monitor intervals | Shorter intervals for near-real-time monitoring |
| Metric granularity | Standard | Finer per-instance measurements |
| Metric naming | Renamed to Aliyun${ProductName}_${MetricName} | Original names preserved |
| Configuration | Automatic with agent V1.1.32+ | Varies by service |
Cloud Monitor metrics
Cloud Monitor exposes metrics for virtually every Alibaba Cloud service. For the full catalog, see Metrics of Alibaba Cloud services. If the service you need is not listed in the Integration Center, submit a ticket.
Automatic synchronization
Starting from agent version V1.1.32 (v5.1-e0c90c68), the Managed Service for Prometheus agent automatically synchronizes Cloud Monitor metrics. When a cloud service reports a new metric to Cloud Monitor, Managed Service for Prometheus picks it up without additional configuration.
Once connected, Cloud Monitor metrics for a service instance are available across all regions.
Troubleshoot missing metrics
If a new Cloud Monitor metric does not appear in query results, verify the agent version:
Log on to the Application Real-Time Monitoring Service (ARMS) console.
Go to Integration Management > Integrated Environments > Cloud Service Region and select the target environment.
On the Configure Agent tab, confirm the agent version is V1.1.32 or later.
If the version meets the requirement, go to Component Management and upgrade any outdated components.
Metric naming rules
After a Cloud Monitor metric is connected, Managed Service for Prometheus renames it to Aliyun${ProductName}_${MetricName} to prevent naming conflicts across services.
Example: The ECS metric CPUUtilization becomes AliyunEcs_CPUUtilization.
The full transformation works as follows:
Step 1 -- Derive MetricName: Replace any special character in the original Cloud Monitor metric name (characters other than letters, digits, and underscores) with an underscore (_).
Step 2 -- Derive ProductName (standard rules): Transform the Cloud Monitor namespace:
Remove the
acs_prefix.Remove the
_dashboardsuffix.Remove any underscores (
_) in the middle.Capitalize the first letter and lowercase the rest.
Worked example: acs_ecs_dashboard -> remove acs_ -> ecs_dashboard -> remove _dashboard -> ecs -> capitalize -> Ecs.
ProductName exceptions
For historical reasons, some services use non-standard product names. The following table lists these exceptions, sorted by namespace.
| Namespace | ProductCategory | ProductName |
|---|---|---|
| acs_amqp | amqp_instance | Amqp_instance |
| acs_bandwidth_package | sharebandwidthpackages | Ebwp |
| acs_cc | cc_iot | Cciot |
| acs_disk | disk_replica_pair_async, dbsc, disk, disk_replica_group_async | Dbsc, Disk |
| acs_elasticsearch | elasticsearch | Es, Elasticsearch |
| acs_elasticsearch | logstash | Logstash |
| acs_emr | emr, emr_serverless_starrocks, emr_workflow | Starrock, Emr |
| acs_fnf | functionflow | Functionflow |
| acs_global_acceleration | acs_global_acceleration, globalaccelerationplus | Ga |
| acs_gws | gws | Ecd |
| acs_hbaseserverless | hbaseserverless | Hbase, Hbaseserverless |
| acs_hitsdb | influxdb, tsdb | Influxdb |
| acs_hologres | hologres_follower | Hologres_follower |
| acs_hologres | hologres_shared | Hologres_shared |
| acs_hologres | hologres_standard | Hologres_standard |
| acs_hologres | hologres_warehouse | Hologres_warehouse |
| acs_hybriddb | gpdb | Gpdb |
| acs_ipv6_bandwidth | ipv6gateway_gw | Gw |
| acs_kafka | kafkaV3, kafka | AliKafka |
| acs_kvstore | kvstore_sharding, kvstore_splitrw, kvstore_standard, kvstore_old | Redis, Kvstore |
| acs_learn | learn_eas | Learn_eas |
| acs_learn | learn_eas_resource | Pai_learn_eas_resource |
| acs_maxcompute_prepay | maxcompute_common | Maxcompute_common |
| acs_maxcompute_prepay | maxcompute_subscription | Maxcompute_subscription |
| acs_maxcompute_prepay | maxcompute_tunnel_subscription | Maxcompute_tunnel_subscription |
| acs_mongodb | mongodb_replicaset | Mongodb_replicaset |
| acs_mongodb | mongodb_singlenode | Mongodb_singlenode |
| acs_nat_gateway | nat_gateway, enhanced_nat_gateway, nat_bandwithpackage | Nat, Natgateway |
| acs_newbgpddos | newbgpddos | ddos-pro |
| acs_pai_dlc | dlc | Paidlc |
| acs_pai_dsw | dsw | Paidsw |
| acs_pai_quota | quota | Paiquota |
| acs_physical_connection | port, vbr | Port, Vbr |
| acs_rds_dashboard | rds | Rds |
| acs_rds_dashboard | rds_cluster | Rds, Rds_cluster |
| acs_rds_dashboard | postgresql | Rds |
| acs_rds_dashboard | sqlserver | Rds, Sqlserver |
| acs_rocketmq | rocketmq | Rocketmq5 |
| acs_vpc_eip | eip | Eip, Vpceip |
| waf | waf, hybrid_waf, wafv3, hybrid_wafv3 | Waf |
Advanced metrics
Advanced metrics are scraped directly by Managed Service for Prometheus instead of being relayed through Cloud Monitor. Compared to Cloud Monitor metrics, they offer:
Finer granularity with more detailed per-instance measurements.
Shorter collection intervals for near-real-time monitoring.
Original metric names preserved as-is, with no
Aliyunprefix added.
Unlike Cloud Monitor metrics, advanced metrics are not connected across all regions automatically. The default provisioning policy depends on the service.
Supported services and provisioning policies
| Provisioning policy | Cloud services |
|---|---|
| Same-region only -- Connects metrics only for instances in the same region as the data storage region | Microservices Engine (MSE), Lindorm, ApsaraMQ for Confluent, ApsaraMQ for Kafka, ApsaraMQ for RocketMQ (5.0), Service Mesh (ASM), PAI-EAS online model service, Intelligent Computing LINGJUN |
| By cluster -- Connects metrics at the cluster level | Container Service for Kubernetes (ACK), Serverless Container Service for Kubernetes (ASK), Container Compute Service (ACS), Distributed Cloud Container Platform for Kubernetes (ACK One) |
| By workspace -- Connects metrics at the workspace level | Realtime Compute for Apache Flink |
Storage
After cloud service metrics are connected, a Managed Service for Prometheus instance is automatically created for data storage. For details on storage configuration and retention, see Data storage.