Application Real-Time Monitoring Service:What is Managed Service for Prometheus?

1. Data collection

Managed Service for Prometheus provides full-stack metric collection capabilities covering client applications, server-side applications, containers, hosts, self-managed components, and cloud services. It also supports integrating self-managed Prometheus data through compatibility with the standard Remote Write protocol of open source Prometheus, and writing custom metrics based on SDKs.

• Client applications and server-side applications: Metric data of client applications monitored by RUM and server-side applications monitoring by APM are written to Managed Service for Prometheus by default. Custom metrics within applications can be implemented via Prometheus SDK or using OpenTelemetry SDK for custom metric ingestion.

• Containers: Alibaba Cloud container services including Container Service for Kubernetes (ACK), ACK Serverless, and Container Compute Service (ACS), integrate with Managed Service for Prometheus by default. The self-developed Prometheus agent enables basic container metric collection, and supports custom metric collection through ServiceMonitor and PodMonitor. For self-managed Kubernetes clusters or Kubernetes services from other cloud service vendors, clusters registered in ACK One can be used for management, enabling container metric collection in the same way as ACK.

• Hosts: For virtual host monitoring, Managed Service for Prometheus supports collecting CPU, memory, disk, network, and other OS-level metrics through the managed Prometheus agent. It also supports collecting process- and container-related metrics. On Elastic Compute Service (ECS), custom metrics can be collected using methods similar to the textfile format of node-exporter. For data centers or virtual host services from other cloud service vendors, deploying open source node-exporter and Prometheus components allows for data collection, which can then be aggregated into Managed Service for Prometheus through Remote Write.

• Cloud services: Managed Service for Prometheus supports integration with CloudMonitor data via the web-based interface, enabling the collection of cloud service metrics. Additionally, the cloud service integration capability automatically enriches cloud service instance names and tags onto the metrics, providing multi-dimensional aggregation, filtering, and alert routing capabilities.

• Custom metrics: Managed Service for Prometheus is compatible with open source Prometheus and OpenTelemetry Metrics protocols, allowing you to implement custom metric ingestion through SDKs.

2. Data storage

Managed Service for Prometheus supports tiered storage capabilities. When standard storage data expires, it can be automatically archived to archive storage, meeting requirements for long-term metric storage and low-cost solutions.

• Managed Service for Prometheus standard storage supports various storage durations, with charges based on data write volume or data reporting volume.

• If you have longer-term storage requirements, you can enable archive storage capabilities. Once enabled, data exceeding the specified duration in standard storage will be automatically moved to archive storage, which is billed based on the storage volume.

3. Data consumption

Managed Service for Prometheus can achieve unified metric data collection and unified storage, meeting the needs of various data consumption scenarios.

• Unified visualization: Managed Service for Prometheus provides rich out-of-the-box dashboards for cloud services and common open source components. You can import built-in dashboards into Managed Service for Grafana or customize dashboards based on Grafana services. Additionally, based on the Prometheus HTTP API, you can also integrate with Alibaba Cloud DataV to build more vivid visualization panels.

• Global view: The aggregated view capability provided by Managed Service for Prometheus supports cross-instance and cross-account Prometheus instance aggregation queries, meeting requirements for unified monitoring across accounts.

• Data processing: Managed Service for Prometheus provides pre-aggregation capabilities through Recording Rules, compatible with open source Recording Rule specifications, offering capabilities such as down-sampling and dimensionality reduction for metric data based on Recording Rules to reduce data costs and improve query performance.

• Unified alerting and response: Managed Service for Prometheus has built-in alerting rules for cloud services and common open source components by default and also supports customizing alert rules based on PromQL. It also supports importing alert rules from open source Prometheus. After alerting rules generate alert events, they are uniformly written into Alert Management, where actions such as alert notifications, assignments, and escalations, can be implemented.

• CloudLens: For core monitoring scenarios, Managed Service for Prometheus provides unified observability based on CloudLens for logs, metrics, and events, meeting comprehensive observability needs for cloud services.

• Data analysis:Managed Service for Prometheus is 100% compatible with PromQL syntax, supporting performing custom analysis and exploration based on PromQL, while also providing metric management capabilities to help you understand the distribution and high cardinality of metrics.

• Data export: Managed Service for Prometheus supports real-time delivery, allowing metric data to be delivered in real time to Kafka, MaxCompute, and self-managed Prometheus services.