We implemented storage sharding in our open-source Postgres multi-tenant storage engine, Pageserver, to support large datasets without competing resources on the same server. This allows us to distribute data across multiple Pageservers, delivering higher storage capacity and throughput, while maintaining invisible configuration for developers. Storage sharding has several advantages, including improved performance for smaller databases, reduced impact of maintenance operations, and increased scalability for larger datasets. The implementation involves dividing data into shards based on a pseudorandom hash function and storing them on separate Pageservers, with the ability to split shards and optimize read and write paths through shard subscription and routing. Our current architecture allows us to scale tenants to tens of terabytes of data and potentially hundreds of terabytes in the future, but we plan to implement further optimizations such as WAL fan-out and shard splitting to further improve performance.