Realtime Compute for Apache Flink supports two monitoring services for Flink job monitoring and alerting: Cloud Monitor and Managed Service for Prometheus of Application Real-Time Monitoring Service (ARMS). Cloud Monitor is free of charge. Compare the two services and choose the one that fits your requirements.
Choose a monitoring service
| If you need... | Use |
|---|---|
| A free, zero-setup monitoring solution | Cloud Monitor |
| PromQL-based queries and Grafana dashboards | Managed Service for Prometheus of ARMS |
| Alert rules for more than 7 metrics | Cloud Monitor (up to 20) or ARMS console (unlimited via PromQL) |
| Scheduled alert notifications | Cloud Monitor or the ARMS console |
| System and custom event-triggered alerts | Cloud Monitor |
| Proactive alerting | Cloud Monitor |
| Longer data retention (90 days) | Managed Service for Prometheus of ARMS |
Feature comparison
| Category | Item | Managed Service for Prometheus of ARMS | Cloud Monitor |
|---|---|---|---|
| Service availability | N/A | ≥99.90% for both monitoring and alerting. See Managed Service for Prometheus Service Level Agreement. | ≥99.9% for alerting only. Availability for the monitoring feature is not guaranteed. See Cloud Monitor Service Level Agreement. |
| Cost | Monitoring and alerting fees | Charged for metric collection and alerting on a pay-as-you-go basis. | No fees for subscriptions, O&M, or metric and alert management. Alert text messages and phone calls are charged by volume at a low price. |
| Data retention | Retention period | Free quota of 50 GB/month. Data retained for 90 days by default. See Managed Service for Prometheus: Instance billing. | No storage fees. Data retained for 30 days by default. |
| Metrics | Display | Displayed in monitoring charts in the Flink development console. Supports PromQL queries and Grafana dashboards in the ARMS console. | Pre-aggregated metrics displayed in the Flink development console and the Cloud Monitor console. For limits compared with ARMS, see the FAQ about the monitoring and alerting features and logs. |
| Number of metrics | In the Flink development console: see Monitoring metrics. In the ARMS console: see Metrics. | See Monitoring metrics. | |
| Alert management | Configuration methods | Configure alert rules in the Flink development console or the ARMS console. | Redirect from the Flink development console to the Cloud Monitor console for configuration or subscription. |
| Number of metrics for alert rules | Up to 7 metrics in the Flink development console (including deployment failure events). See Configure alert rules. Unlimited metrics in the ARMS console via PromQL. Subscribe to Elastic Compute Service (ECS) failure handling events and ECS proactive O&M events in the Cloud Monitor console. See Configure alert rules. | Up to 20 metrics. Supports event-triggered alerts for deployment failure events, ECS failure handling events, and ECS proactive O&M events. | |
| Single or multiple metrics | Single metric per rule in the Flink development console (metric can be deployment failure events). Multiple metrics per rule in the ARMS console via PromQL. | One or more metrics per rule. Subscribe to alerts for deployment failure events. | |
| Alert rules | Create in the Flink development console or in the ARMS console. | Create in the Cloud Monitor console. | |
| Alert templates | Create in the Flink development console or in the ARMS console. | Create in the Cloud Monitor console. | |
| Event-triggered alerts | Supports only deployment failure events. System and custom event-triggered alerts are not supported. | Supports system event-triggered alerts and custom event-triggered alerts. Alerts can be delivered to a Simple Message Queue (SMQ, formerly MNS) queue, a specified function, an Internet-accessible URL, or Simple Log Service (SLS). | |
| Alert notification | Notification methods | In the Flink development console: phone calls, text messages, emails, DingTalk, and webhook. Scheduling is not supported. In the ARMS console: phone calls, text messages, emails, DingTalk, WeCom, Lark, and webhook. Scheduling is supported. See Notification objects. | Phone calls, text messages, emails, DingTalk, WeCom, Lark, and webhook. Scheduling is supported. See Alert contacts. |
| Notification policies | In the Flink development console: alert denoising policies. In the ARMS console: alert silence policies and alert escalation policies. See Create and manage an escalation policy. | Dynamic threshold-triggered alerting, merged alert notifications, and blacklist policies. See Dynamic threshold-triggered alert rules, Merge alert notifications, and Manage blacklist policies. | |
| Alert callback | Supported. | Supported. | |
| Proactive alerting | Not supported. | Supported. | |
| API operations | Monitoring and alerting APIs | List of operations by function and Obtain remote read, remote write, and HTTP API endpoints. For Flink metrics, see Flink metrics. | API operations for cloud service monitoring and API operations for alerting. |
Switch the monitoring service for a workspace
To switch the monitoring service for your workspace, log on to the Flink Management Console and click 
in the Actions column of your workspace.
Read the prompts carefully before confirming the change.
For details, see Configure alert rules.