DataStorm
Distributed database

A distributed database that speaks SQL and pushes every change to you as it happens.

DataStorm speaks PostgreSQL, so psql and your existing drivers connect directly. It also has a chained document query language, and any query can become a live feed of changes. Each table can be stored as rows for transactions or as columns for analytics.

Available 11 October 2026
psql session against DataStorm with a vector similarity query

DataStorm is in early release. Cross-shard transactions use snapshot isolation.

PostgreSQL protocolpsql and libpq drivers work as they are
Live queriesSubscribe to any query and get each change
Rows or columnsChoose storage per table
9 SDKsEach a single file with no dependencies

What DataStorm does

For developers building real-time apps, teams that want one engine for transactions, analytics, search and vectors, and builders of AI agents.

Store and query

  • Two query languages

    Standard SQL and a chained document language, on the same data.

  • Analytics tables

    Columnar tables answered filter queries 27 to 260 times faster than row tables in our tests.

  • Vector search

    Vector columns, HNSW indexes and similarity search combined with ordinary filters.

  • Search, maps and time

    Full-text search, geospatial functions and time-bucketed aggregation.

Analytic queries over a million order lines in a columnar DataStorm table

Real time

  • Changefeeds

    Subscribe to a query and receive every change the moment it commits.

  • Memory for AI agents

    Built-in commands for agents to remember, recall and verify facts, with access per agent.

A DataStorm changefeed receiving rows as they are inserted

Trust and scale

  • Transactions

    Snapshot isolation, savepoints and transactions across shards.

  • Consensus and sharding

    Every shard is replicated with Raft and splits automatically as it grows.

  • Hot standby

    A signed, chained replication log to a second cluster, promoted by an operator when needed.

  • Point-in-time restore

    Continuous log shipping and snapshots to S3 or any other target.

Secure and observable

  • Access control

    Mutual TLS, row-level security and permissions per operation and key prefix.

  • Integrity checks

    Background scrubbing finds silent data corruption.

  • Audit and metrics

    An audit log, Prometheus metrics and storage usage per tenant.

Part of the Vault56 family

DataStorm is a standalone database in the Vault56 family. It runs next to your applications today, and a direct link to the Vault56 platform is on the roadmap.

How the platform works

Get DataStorm on 11 October 2026

Try the platform today. The products open to everyone on 11 October 2026.