Citus 12: Schema-based sharding for PostgreSQL

published on 2023/07/19

Schema-based sharding means that tables from the same schema are placed on the same node, while different schemas may be on different nodes. That way, queries and transactions that involve a single schema can always be evaluated efficiently by a single node (read: without network overhead), while the system can transparently scale out to accommodate an arbitrarily large amount of data and high rate of queries across different schemas. Moreover, the cluster can be rebalanced based on disk usage, such that large schemas automatically get more resources dedicated to them, while small schemas are efficiently packed together.

Citus Data