ZebraDB
Transactions and live analytics on the same operational data.
What it does.
ZebraDB keeps writes in Postgres and makes the same data available for analytical queries without waiting for a batch ETL job. Z-Tables add tenant isolation, history, policy, and lineage. SQL and Arrow keep the database open to applications, agents, and analytics.
Write once. Query the live state and its history.
ZebraDB keeps operational writes in native Postgres while projecting history into an analytical plane. A replication watermark lets DataFusion combine the columnar base with the unreplicated transactional tail.
Command, agent operation, or integration mutation.
Transaction, schema, tenancy, RLS, indexes, and attribution.
Append-only change record with ordered event identity.
Replayable stream into Parquet or a column store.
One logical query through Postgres wire, Arrow, FlightSQL, DOSS Grid, or MCP.
One fresh typed result without exposing physical storage topology.
Plans, latency, lag, and errors drive partitioning, caching, and engine choice.
What it owns
Live HTAP
Read the analytical base and the newest transactional changes in one query.
Z-Tables
Use real Postgres tables with indexes, partitions, tenant isolation, and append-only history.
Open interfaces
Connect through Postgres wire, SQL, Arrow, FlightSQL, DOSS Grid, or MCP.
Every write has an actor
Policy, lineage, history, and attribution stay attached to the operation.
What it connects to
ZebraDB calls these products and services directly.
