Scale a Hyperscale (Citus) server group

Azure Database for PostgreSQL - Hyperscale (Citus) provides self-service scaling to deal with increased load. The Azure portal makes it easy to add new worker nodes, and to increase the vCores of existing nodes.

Add worker nodes

To add nodes, go to the Configure tab in your Hyperscale (Citus) server group. Dragging the slider for Worker node count changes the value.

Resource sliders

Click the Save button to make the changed value take effect.

Note

Once increased and saved, the number of worker nodes cannot be decreased using the slider.

Rebalance shards

To take advantage of newly added nodes you must rebalance distributed table shards, which means moving some shards from existing nodes to the new ones. First verify that the new workers have successfully finished provisioning. Then start the shard rebalancer, by connecting to the cluster coordinator node with psql and running:

SELECT rebalance_table_shards('distributed_table_name');

The rebalance_table_shards function rebalances all tables in the colocation group of the table named in its argument. Thus you do not have to call the function for every distributed table, just call it on a representative table from each colocation group.

Increase vCores

In addition to adding new nodes, you can increase the capabilities of existing nodes. This feature is currently in preview — to request increased vCores for nodes in your server group, please contact Azure support.

Next steps

Learn more about server group performance options.