This topic describes the default limits and quotas for Managed Service for Prometheus V2 instances.
V2 instances store data in the Metricstores of your Simple Log Service projects. This storage model took effect at 00:00:00 (UTC+8) on March 12, 2025. For details, see Storage upgrade of Managed Service for Prometheus instances. For V1 instance limits, see Service limits (legacy, not applicable to Prometheus V2 instances).
Data read and write
For details, see Data read and write in Simple Log Service.
Query and analysis
For details, see Simple Log Service Metricstores.
Data collection and reporting
These limits govern Prometheus agent data collection. Resource-based limits assume a 3-core CPU and 4 GB memory per agent replica unless otherwise noted.
Item | Default limit | Adjustable | Description |
Data points per collection per agent replica | 3,500,000 | Yes | Total data points a single agent replica can collect per scrape cycle. Scale out replicas to increase capacity. |
Targets per agent replica | 5,000 | Yes | Scrape targets assigned to a single agent replica. Scale out replicas to increase capacity. |
Data points per target (30-second window) | 600,000 | Yes | Data points collected from a single target within one 30-second scrape interval. Scale out replicas to distribute the load. |
Replicas in agent HPA settings | 20 | Yes | Agent replicas allowed through Horizontal Pod Autoscaler (HPA). To exceed this limit, manually scale out replicas in the Managed Service for Prometheus console. |
Data volume per target | 512 MB | No | Maximum data volume from a single scrape target. Excess data is discarded. |
Container Service for Kubernetes (ACK) cluster compatibility | Agents V3.2.1 or later require Kubernetes V1.16 or later | Yes | Update your ACK cluster version to meet the minimum Kubernetes version required by the agent. |
Others
Item | Default limit | Adjustable | Description |
Recording rules per cluster | 1,000 | No hard limit, but processing delays may occur. | No hard cap is enforced, but exceeding 1,000 recording rules may cause processing delays. |
Folders per tenant in Managed Service for Grafana Shared Edition | 250 | No | To use more folders, upgrade to Managed Service for Grafana Pro Edition. |