Definition

What it is

Lakekeeper is an open-source (Apache-2.0), **Rust-native Apache Iceberg REST Catalog**. Where the incumbent catalogs grew out of JVM/Hive-metastore lineage, Lakekeeper scales horizontally with no garbage-collection overhead — but its defining feature is **embedded Cedar policy enforcement**: it evaluates user *and agent* identities against declarative ABAC/RBAC policies and vends **short-lived, remote-signed storage credentials**, so data authorization cannot be bypassed by a rogue compute engine or a compromised agent.

Why it exists

Lakekeeper is a control plane over [Apache Iceberg](/node/apache-iceberg) tables on object storage — it sits in the [Iceberg REST Catalog](/node/iceberg-rest-catalog-spec) convergence alongside [Apache Polaris](/node/apache-polaris), [Unity Catalog](/node/unity-catalog), [Apache Gravitino](/node/apache-gravitino), and Nessie. Its differentiator is being the **security catalog** of the agentic data control plane: when the querying identity is an autonomous agent, per-request policy evaluation and scoped credential vending are what keep an agent from reading or mutating data it shouldn't.

Primary use cases

central Iceberg REST catalog for multi-engine lakehouses (Spark, Trino, PyIceberg, Flink); identity-aware access control for agents and engines; short-lived scoped storage-credential vending; self-hosted, GC-free catalog at scale.

Recent developments

Latest signals

v0.12: audit events, batch authorization, and the Generic Table API extend the security-catalog thesis. 0.12.0 ships dedicated audit events with exactly-once-per-call delivery and structured JSON logging (the compliance trail regulated deployments need), batch-authorization optimization on the Cedar/OPA path, and a Generic Table API that registers external formats — Lance and Delta — as first-class generic tables, extending credential vending, soft-delete protection, and authorization policy across formats without faking Iceberg metadata or migrating data. Storage I/O also moved off OpenDAL to hyperscaler-native backend routing, so every byte passes through the vended-credential validation layer. Per Lakekeeper release notes.
Cedar policy enforcement in the catalog (v0.12.x). Lakekeeper integrates AWS's open-source Cedar policy engine directly into the Iceberg REST catalog, evaluating ABAC/RBAC rules per request and issuing short-lived remote-signed credentials — the pattern the 2026 "agentic data control plane" blueprint centers its security layer on. Per Lakekeeper Cedar authorization docs and Lakekeeper (GitHub).
Rust-native, horizontally scalable REST catalog. No JVM, no GC pauses; positioned as the lightweight, high-concurrency alternative for lakehouses standardizing on the Iceberg REST spec. Per Lakekeeper docs and building a modern lakehouse with Lakekeeper + Trino (OLake).
Lakekeeper is building toward the same /metrics telemetry layer as Polaris. Cedar-based authorization already scopes who can read or write; the emerging metrics-persistence work extends that same request path to record ScanReport/CommitReport telemetry per call, fusing Lakekeeper's access-control decision with the operational record of what was actually read. See Iceberg REST Metrics API. Per Lakekeeper Management API docs. Sources: Lakekeeper release notes · Lakekeeper docs · Lakekeeper Cedar authorization · Lakekeeper (GitHub) · Modern lakehouse with Lakekeeper + Iceberg + Trino (OLake)