Find your next open-source project

Search by name, purpose, or keyword — or paste a GitHub repository URL

apache/polaris

Data

An open-source REST catalog for Apache Iceberg that lets Spark, Flink, Trino and other engines share lakehouse metadata.

2,122
575
Java
Apache-2.0
2 years old

Overview

Apache Polaris is an open-source catalog for Apache Iceberg. It implements Iceberg's REST API, so multiple engines — Apache Spark, Apache Flink, Trino, Apache Doris, StarRocks and Dremio OSS among them — can interoperate over the same lakehouse metadata instead of each engine keeping separate table definitions. It targets data platform engineers and lakehouse operators who need one catalog across engines.

The codebase is organized as Gradle modules: polaris-core holds entity definitions and business logic, generated API modules serve the Management API and Iceberg REST service, polaris-server is a Quarkus-based runtime, and Hive/Hadoop/BigQuery federation plus OPA/Ranger authorization are extensions project structure. A Python client and a Spark plugin ship in the repository, with additional tooling in the separate Polaris-Tools repository.

Quickstart

Building requires Java 21+ and Docker 27+, and Docker must be running because integration tests depend on it build instructions.

./gradlew build
./gradlew run -Dpolaris.bootstrap.credentials=POLARIS,root,secret

The server is then reachable at localhost:8181. The bootstrap argument sets realm POLARIS, client ID root and secret secret; without it, the preset credentials are POLARIS,root,s3cr3t. To exercise the catalog, run ./regtests/run_spark_sql.sh and execute, for example:

create database db1;
create table db1.table1 (id int, name string);
insert into db1.table1 values (1, 'a');
select * from db1.table1;

A published image can be started directly with docker run -p 8181:8181 -p 8182:8182 apache/polaris:latest. The Quickstart covers local and cloud setups, and the Overview explains how the catalog works.

Deployment and configuration

Kubernetes deployments use the Helm charts under helm/, documented in the Helm README. Server configuration supports several mechanisms; defaults live in runtime/defaults/src/main/resources/application.properties, and the Configuration Guide documents the options. Docker-dependent integration tests can be skipped with ./gradlew check -PnoIntegrationTests.

Constraints

  • Java 21+ and Docker 27+ are required for the standard build and test flow; integration tests need a running Docker daemon.
  • The ./gradlew run mode falls back to preset credentials when none are supplied, so set your own before exposing the server beyond local development.
  • OPA and Ranger authorization integrations are extensions; the default runtime configuration should be reviewed before production use.
  • The project is licensed under Apache-2.0 license.

HotRepo's take

On 2026-11-19, Apache Polaris reached 9th place on the Java daily list with 2,121 stars and 575 forks, recording no star gain in that snapshot. That single record places it in Java's daily top ten, but it does not establish a longer trend. The more telling fact is its role as an implementation of Iceberg's REST catalog rather than a single-engine utility: its usefulness depends on adoption by the surrounding engines, which is why the federation, authorization and client modules carry much of the practical weight.

Written by AI from the project's README, updated . The project's own docs are the source of truth.

You might also like

  1. apache/icebergA table format that lets several query engines read and write the same huge analytic tables concurrently.Java9,379
  2. apache/gravitinoUnifies metadata across Hive, MySQL, S3 and Iceberg-style lakehouses so query engines and teams see one catalog instead of many.Java3,322
  3. datazip-inc/olake-goData ingestion engine syncing databases, Kafka and S3 into Apache Iceberg or Parquet for lakehouse pipelines.Go2,007
  4. apache/dorisA real-time SQL analytics database for sub-second dashboards, lakehouse queries, and hybrid text/vector search.Java16,160
  5. apache/tikaDetects and extracts metadata and text from over a thousand file types for Java applications and AI/RAG pipelines.Java4,128