Yandex Managed Service for Sharded PostgreSQL

No. The service provides a single PostgreSQL-compatible endpoint so that your application continues to run without changes.

Yes. For example, with Yandex Managed Service for PostgreSQL.

Sharded PostgreSQL is not a replacement, but the next stage of evolution for PostgreSQL, when load growth requires horizontal scaling. The choice depends on your application architecture and load type.

A sharded PostgreSQL is better for the following scenarios:

• Monolith performance issues: When you hit single instance limits (CPU, disk, network), vertical scaling doesn’t help anymore or becomes too expensive. Sharded PostgreSQL allows you to increase resources by adding new shards.
• Large amounts of data: when the database size exceeds several terabytes and does not fit on one machine.
• High OLTP loads: Sharded PostgreSQL is optimized for short transactions with high query rates, where user data is logically isolated within a single shard (for example, all of one user’s operations take place in his shard).
• Development from scratch: it is much easier to implement sharding on the new service (the implementation period is ~1 month).
• Manage multiple instances: Shaded PostgreSQL makes it easy to manage multiple small PostgreSQL instances by offering them to the application as a whole.

Regular PostgreSQL is better in these scenarios:

• Complex cross-shard queries: if your application actively uses JOIN between tables that will be located on different shards. Sharded PostgreSQL does not support such queries.
• Complex distributed transactions: Query chains involving multiple shards only supported on a limited basis or via 2PC, but this is not Sharded PostgreSQL’s strong suit. Regular PostgreSQL handles this more reliably.
• Analytical loads (OLAP): Sharded PostgreSQL is not a substitute for analytical databases like Greenplum^®. If you need heavy analytical queries on all your data, Sharded PostgreSQL is not suitable.
• Difficult migration of an existing application: Migrating an already running monolithic application to Sharded PostgreSQL requires logic improvements (for example, allocating a sharding key) and can take considerable time (from 3 to 6 months, sometimes more).

Yandex Managed Service for Sharded PostgreSQL operates at the PostgreSQL protocol level. This assumes full compatibility with different versions. As of early 2026, these are versions 14, 15, 16, 17, and 18.

Yes. The service is fully compatible with PostgreSQL data types, including JSONB.

You can find out the pricing rules for the service in the documentation.

Yes, the support team helps users of Yandex Managed Service for Sharded PostgreSQL just like other cloud clients.