Docs

Volcengine Managed Service for Prometheus

Product overview

Product advantages

Quick cloud service integration

Volcengine Managed Service for Prometheus (VMP) allows you to quickly create standard Prometheus instances on the cloud without manual configuration and tuning. It provides a one-stop service integration center to simplify integration. VMP supports configuration with a few clicks, which quickly collects metrics from AI frameworks, open-source plugins, Volcano Engine cloud services, and multi-cloud/hybrid cloud services.

VMP can be deeply integrated with AI-related services of Volcano Engine to monitor the basic cloud resources of Volcano Engine, large model platforms, and self-built large AI models such as Doubao and DeepSeek. The services include Machine Learning Platform (MLP), ModelArk, Volcengine Kubernetes Engine (VKE), Elastic Compute Service (ECS), and other services. When you deploy private large AI model services on Volcano Engine, you can achieve full-stack large AI model monitoring for basic resources (such as CPU and GPU), storage, network (such as QPS and response time), and call tokens.

Cloud services of Volcano Engine can be integrated with VMP through direct integration with Prometheus workspaces or through integration with Cloud Monitor. This enables you to build a one-stop monitoring platform for basic resources, middleware, storage, network, and custom businesses.

VMP supports cross-site monitoring and cross-region integration. VMP allows you to connect networks across different vendors, environments, or regions through Direct Connect (DC), Cloud Enterprise Network (CEN), or other services. This achieves cross-site and cross-region monitoring in a centralized manner. In addition, VMP allows you to use local workspaces, aggregation query workspaces, and federated query endpoints to aggregate and query hierarchically federated, multi-form Prometheus data.

Various preset dashboards of agents

VMP provides preset dashboards of AI frameworks, open source agents, and cloud service metrics to enable you to quickly build monitoring dashboards.

Multi-replica high availability

You can use the high-availability solution that deploys multiple replicas in a single availability zone and across multiple zones. The solution ensures that VMP is running as expected in case of instance failure or availability zone failure.

Zero construction and O&M

As a fully managed service, VMP can save you on usage costs compared with self-built Prometheus.

The fully managed service saves you from considering storage and O&M costs for large-scale data.

VMP supports billing methods such as pay-as-you-go and package, and offers a large number of metrics free of charge for common instance specifications.

No manual construction or O&M is required, which saves manpower.

Open source ecosystem affinity

VMP is a service that is compatible with the open source ecosystem, which is reflected in the following aspects:

VMP is compatible with the Prometheus community ecosystem. VMP supports third-party exporters provided by the Prometheus community, which can meet the monitoring requirements for various environments and agents of users.

VMP is compatible with the configuration files and configuration methods of native Prometheus.yaml collection rules. This allows you to configure custom collection rules.

VMP is compatible with PromQL syntax.

Therefore, VMP facilitates migration. Users who are using native Prometheus can directly migrate configuration files, code integrated with native Prometheus, and Grafana dashboards to VMP.

Last updated: 2026.07.10 11:37:27