Resource quotas let you partition purchased resources across teams and projects, enforce usage limits per workspace, and control how idle capacity is shared or preempted — keeping your cluster fully utilized without contention.
Resource quota types
After purchasing resources in the resource pool, create resource quotas or adjust the limits of existing ones. Resources from different resource groups can share the same resource quota, and you can build a multi-level hierarchy of child resource quotas to reflect your organizational structure.
Two quota types are supported:
-
Cloud-native resource quotas: cover Lingjun resources and general computing resources. Quotas form a tree structure called QuotaTree — the Root Quota draws from the resource pool, and you subdivide it into one or more levels of Child Quotas. This hierarchy provides enhanced queueing and scheduling capabilities, letting you allocate capacity by organization or project and manage jobs more efficiently.
The resource quota created by allocating resources from the resource pool is the Root Quota.
-
With the Root Quota as the parent node, you can create multiple levels of Child Quotas.
Divide and assign quotas by organization or project to manage jobs and resources more efficiently.
Big data engine resource quotas: cover MaxCompute and fully managed Flink resources.
Create and use resource quotas
-
Cloud-native resource quotas
To create Lingjun resources or general computing resource quotas, see Create a resource quota.
-
Big data resource quotas
For MaxCompute resources, see MaxCompute resource quota.
For fully managed Flink resources, see Fully managed Flink Resource Management.
Manage cloud-native resource quotas
Basic configurations: create child-level quotas, scale quota limits, and view quota lists and details. For more information, see Manage resource quotas.
-
Advanced configurations:
Scheduling policy: Control dequeuing order and resource utilization by choosing a scheduling policy.
Preemption policy: Allow tasks in the current resource quota to reclaim computing resources from peer or child-level quotas, keeping overall utilization high.
Resource quota monitoring and alerting: View usage metrics, set alert thresholds, and subscribe to monitoring notifications via Cloud Monitor and ARMS.
To allocate resources across teams using parent-child quota hierarchies, see Training and inference resource management.