Iceberg Catalog Interface

Run Apache Iceberg without external catalog infrastructure. AIStor embeds the Iceberg REST Catalog API directly into object storage.

Get a Demo

Diagram showing Spark, Trino, and Flink connecting via REST API to AI STOR with a note below stating External Catalog Not Required.

Native Iceberg REST Catalog

Full Apache Iceberg REST Catalog API built directly into AIStor—no external catalog services required.

REST Catalog API

Fully compliant Iceberg catalog REST endpoints for catalog, table, and metadata operations.

Query Engine Integration

Connect Spark, Trino, PyIceberg, Flink and others using the the standard Iceberg REST Catalog API.

Atomic Transactions

Commit changes across tables in a single metadata swap. Every reader sees the full previous or full new state, never partial writes.

Time Travel

Query any previous table state by snapshot ID or timestamp from the immutable snapshot metadata tree.

Schema Evolution

Add, rename, drop, or widen columns in metadata only. Iceberg tracks columns by unique ID, not name.

Partition Evolution

Update partition layout on new writes only. Iceberg records which spec applies to each data file.

Why AIStor is Different

Traditional Iceberg deployments require a separate transactional database like PostgreSQL, MYSQL or MongoDB for catalog metadata. This additional layer adds the operational complexity securing, patching, and scaling an additional service. AIStor Tables eliminates that dependency entirely.

Zero External Catalogs

No separate catalog infrastructure to manage

Flexible Compute Layer

Connect any Iceberg-compatible engine such as Spark, Trino, Flink, or PyIceberg.

Unified Data Store

Manage structured Iceberg tables and unstructured object data from a single platform.

Zero-Config Tables

Deploy AIStor and start creating tables. No additional catalog infrastructure to provision, configure, or manage.

Business Impact

Infrastructure Consolidation

One vendor for unstructured and structured data - fewer components to deploy, monitor, and maintain

Security Posture

One platform, one identity model. Tables and objects share the same encryption, access policies, and audit trail.

Durability and Resilience

Tables inherit AIStor's erasure coding and bitrot protection with both catalog metadata and data distributed across available nodes for hardware fault tolerance.