Manage automatic load balancing

Load balancing for partition replicas

OceanBase Database provides a parameter for you to manage load balancing for partition replicas.

The enable_rebalance parameter specifies to enable or disable load balancing for resource units and partition replicas. When the value of enable_rebalance is False, load balancing is disabled for partition replicas. When the value is True, this feature is enabled.

Load balancing for leaders

OceanBase Database provides a parameter for you to manage load balancing for leaders.

The enable_auto_leader_switch parameter specifies to enable or disable load balancing for leaders. When the value of enable_auto_leader_switch is False, load balancing is disabled for leaders. When the value is True, this feature is enabled.

Control resource unit balancing

OceanBase Database provides three parameters for you to manage load balancing for resource units.

• The enable_rebalance parameter specifies to enable or disable load balancing for resource units and partition replicas. When the value of enable_rebalance is False, load balancing is disabled for resource units and partition replicas. When the value is True, load balancing for resource units is enabled or disabled based on the value of the resource_soft_limit parameter.

• The resource_soft_limit parameter specifies to enable or disable load balancing for resource units. When the value of enable_rebalance is True, load balancing is enabled or disabled for resource units based on the value of this parameter. When the value of resource_soft_limit is less than 100, load balancing is enabled for resource units. When the value is greater than or equal to 100, this feature is disabled.

• The server_balance_cpu_mem_tolerance_percent parameter specifies the threshold to trigger load balancing for resource units. When the difference between the resource load of one or more OBServers and the average resource load of all OBServers in a cluster reaches the value that you specified for the server_balance_cpu_mem_tolerance_percent parameter, load balancing is scheduled. The scheduling for load balancing stops when the difference is less than the value specified for the server_balance_cpu_mem_tolerance_percent parameter.