Cluster Scale

It controls the maximum number of worker nodes that a cluster can manage. Configure the parameter as required. After the cluster is created, it can only be scaled out.

• s1: specifies a CCE cluster with one master node.
• s2: specifies a CCE cluster with multiple master nodes.
• Maximum number of worker nodes: The maximum number of worker nodes in a cluster. Select a cluster flavor based on your requirements.
• A CCE cluster with one master node: If the master node is down, the cluster will become unavailable, but running workloads will not be affected.
• A cluster with multiple master nodes: This is an HA cluster. If a master node is down, the cluster can still be functionally available.

Value Range

• cce.s1.small: a small-scale CCE cluster with one master node and a maximum of 50 worker nodes
• cce.s1.medium: a medium-scale CCE cluster with one master node and a maximum of 200 worker nodes
• cce.s2.small: a small-scale CCE cluster with multiple master nodes and a maximum of 50 worker nodes
• cce.s2.medium: a medium-scale CCE cluster with multiple master nodes and a maximum of 200 worker nodes
• cce.s2.large: a large-scale CCE cluster with multiple master nodes and a maximum of 1,000 worker nodes
• cce.s2.xlarge: an ultra-large-scale CCE cluster with multiple master nodes and a maximum of 2,000 worker nodes

Default Value

cce.s1.small

Modifiable

You can modify cluster specifications to have a different cluster flavor.

Scope

CCE standard and CCE Turbo clusters

Suggestions

Configure clusters with one master node for trial use of these clusters. For scenarios that require high availability, select clusters with multiple master nodes.

Select a suitable cluster flavor based on the maximum predicted capacity of the service and reserve a certain buffer. For example, if 50 worker nodes are about to be used, configure a cluster with 200 worker nodes. When the cluster is in use, monitor the resource usage on the cluster management plane. If the resource usage is high, adjust the cluster flavor in a timely manner to prevent the cluster from being overloaded.