Iceberg REST Catalog Comparison: Polaris, Gravitino, Unity

Iceberg REST Catalog Comparison: Polaris, Gravitino, Unity

Polaris, Nessie, Gravitino, Unity Catalog, Glue, and BigLake compared on RBAC, credential vending, federation, audit, and where each catalog can run.

By

Billy Allocca

Table of Contents

Last updated: November 17, 2026

The Iceberg catalog decides who controls access, credentials, and which engines may write to your tables. Apache Polaris, Project Nessie, Apache Gravitino, Unity Catalog, AWS Glue, and Google’s BigLake (now the Lakehouse runtime catalog) differ on role-based access, credential vending, federation of other catalogs, audit, and where they can run. This guide compares the six on those criteria.

The governance work a catalog absorbs is the work most enterprises repeat today. In NexusOne’s 2026 State of the Data Estate report, conducted by Enterprise Management Associates (EMA) in September 2026 (n=202) and commissioned by NexusOne, the respondents who implement at least some policies separately per system (n=146) most often reimplement access permissions (67%) and classification or sensitive-data tagging (57%) [1]. In NexusOne’s reading, a shared catalog policy layer is where those two belong. This guide is the decision-level companion to the broader Iceberg metadata governance guide, and it goes deep on one question: which catalog, judged on governance.

Why Does the Catalog Decide Whether Iceberg Stays Open?

The table format is open by specification, but every commit and every credential passes through the catalog, so the catalog is where openness is either kept or traded away. A catalog is the service that maps a table name to its current metadata file and accepts commits that move that pointer. An Iceberg REST catalog is a catalog that speaks the HTTP protocol defined in the Apache Iceberg OpenAPI specification, so any compliant engine (Spark, Trino, Flink, DuckDB) can use it without a vendor SDK [2].

Three governance powers concentrate on that commit path. The catalog authenticates the engine, it decides whether that principal may read or write a namespace or table, and it can hand the engine short-lived storage credentials so nobody holds long-lived bucket keys. Credential vending is that last step: the catalog returns table metadata together with downscoped, temporary storage credentials scoped to the table’s files [2][31]. If those three powers live in one cloud’s identity system, the Parquet files stay portable but the policy that protects them does not travel. The format-level argument is in the earlier post on why Iceberg is becoming the default storage format for enterprise AI.

Use these criteria to judge any catalog on governance:

Criterion

What to check

Why it matters

Access model

Role hierarchy, privilege granularity (catalog, namespace, table, view), deny rules, external policy engine support

Decides whether one grant can express your policy or you maintain parallel rules per engine

Credential vending

vended-credentials and remote-signing support, storage types covered, refresh

Removes static bucket keys from engines and agents

Multi-engine writes

Which engines can create and commit through the REST endpoint, and with what limits

A read-only catalog for outside engines forces one writing engine

External catalog federation

Which catalogs it can mount (Hive Metastore, Glue, other REST catalogs) and whether federated tables are writable

Determines whether you can govern existing estates before moving tables

Audit

What is logged (authentication, authorization decisions, commits) and where it lands

Evidence for every access decision an auditor asks about

Deployability

Self-hosted, on-prem, air-gapped, or managed in one cloud only

Decides whether the same catalog can govern your data center and your clouds

License

Apache 2.0, other open license, or proprietary service

Decides who controls the roadmap and the exit

What Does the REST Catalog Specification Standardize, and What Does It Leave to Vendors?

The specification standardizes how engines talk to a catalog, and it leaves the permission model entirely to each implementation. The Iceberg REST Catalog API (OpenAPI 3.1.1, “the first version of the REST Catalog API”) defines endpoints for configuration, namespaces, tables, views, multi-table transactions, scan planning, metrics reporting, and a per-table loadCredentials call [2]. An optional request header, X-Iceberg-Access-Delegation, lets a client declare that it accepts vended-credentials, remote-signing, or both [2]. Remote signing is the alternative to vending in which the catalog signs each storage request for the engine, so credentials never leave the catalog at all.

The specification contains no grant, role, or policy endpoint. It configures routes with an Authorization header and returns 403 when a request is forbidden [2], but how the server decides is up to the vendor. That gap is why six catalogs that all “support Iceberg REST” can govern the same table in six incompatible ways.

Area

Standardized in the REST spec

Left to each catalog

Authentication

Authorization header with OAuth2 or bearer token; the old /v1/oauth/tokens route is deprecated for removal in Iceberg 2.0 [2]

Which identity provider issues tokens and how principals map to roles

Authorization

403 response code [2]

Roles, privileges, inheritance, deny rules, row and column policy

Storage access

Access-delegation header and loadCredentials endpoint [2]

Which clouds and object stores can be vended, credential lifetime, refresh

Transactions

transactions/commit for atomic multi-table commits [2]

Whether the server implements it

Views

View endpoints, including register-view [2]

Whether the server implements them

Federation

Nothing

Mounting Hive Metastore, Glue, or other REST catalogs

Audit

Metrics reporting endpoint [2]

Audit log format, scope, and retention

Implementation coverage varies too. In one September 2026 practitioner test of seven anonymized managed catalogs, eleven endpoints were declared in configuration but not served across four vendors, and view support was typically absent; the author works at Dremio, which ships a Polaris-based catalog, so treat it as an early signal [37]. Test the endpoints you depend on against the catalog you choose.

How Do Polaris, Nessie, Gravitino, Unity Catalog, Glue, and BigLake Compare on Governance?

The three Apache-licensed catalogs run anywhere and differ in what they govern best, while the three cloud catalogs govern most deeply inside their own cloud and none runs in your data center. The table uses each project’s documentation as of October 2026.

Catalog (status)

Access model

Credential vending

Multi-engine writes via REST

External catalog federation

Audit

Runs on-prem

License

Apache Polaris (ASF top-level project Feb 2026; 1.8.0, Sep 28, 2026) [3][4]

RBAC: principal roles hold catalog roles that carry privileges such as TABLE_READ_DATA and TABLE_WRITE_DATA; OPA external authorization in preview [5][6]

Yes, read-only or read-write credentials tied to the privilege [5]

Yes for internal catalogs [7]

Iceberg REST, Hive Metastore (optional build, single HMS), BigQuery metastore [8][9]

Event listeners for authentication, authorization, and catalog changes [10]

Yes, self-hosted

Apache 2.0

Project Nessie (pre-1.0 release line) [13]

CEL rules on reference (branch or tag) and path; governs metadata only [12]

Vended credentials or S3 request signing; GCS and ADLS experimental [11]

Yes, but Nessie labels its Iceberg REST support experimental [11]

None documented; check current documentation

Check current documentation

Yes, self-hosted

Apache 2.0 [14]

Apache Gravitino (ASF top-level project Jun 2025; 1.3.x) [15][16]

Metalake RBAC with users, groups, roles, ALLOW and DENY privileges, deny by default; pushdown to Apache Ranger [18][19]

S3, GCS, OSS, ADLS through its Iceberg REST service; refresh added in 1.3.0 [16][17]

Yes

Hive, JDBC, and REST backends behind one Iceberg REST endpoint; federated REST support in 1.3.0 [16][17]

Pluggable audit log, JSON formatter in 1.3.0 [16][17]

Yes, self-hosted

Apache 2.0 [15]

Unity Catalog, Databricks managed [21][22]

Unity Catalog grants plus EXTERNAL USE SCHEMA for outside engines [22]

Yes; not for tables with row filters or column masks [22]

Read, write, and create for managed Iceberg tables from Spark, Flink, Trino, Snowflake [21]

Hive metastores, Glue, Snowflake, Salesforce Data 360, Palantir Foundry, Workday; external ones read-only [23]

system.access.audit system table [24]

No (AWS, Azure, Google Cloud); OSS edition self-hosted

Proprietary; OSS 0.5.x under Apache 2.0 [20][25]

AWS Glue Data Catalog [26]

IAM plus Lake Formation; third-party credential vending off by default [29]

Through Lake Formation application integration [29]

Create and update tables; no view APIs, single-level namespaces, 50 MB metadata cap [26][27]

Remote Iceberg REST catalogs, GA Nov 24, 2025 [28]

AWS CloudTrail, with some responses logged as null [30]

No (AWS regions)

Proprietary

BigLake / Lakehouse runtime catalog (GA Nov 19, 2025) [33][34]

IAM roles; table-level IAM in preview; no row- or column-level security on these tables [31][34]

Yes, downscoped storage tokens [32]

Spark and Trino; BigQuery DDL and DML cannot create or modify these tables; Parquet only [31]

Unity Catalog on AWS, Glue, Snowflake Horizon, Workday (all preview) [34]

Check current documentation

No (Google Cloud; regional endpoints for residency) [31]

Proprietary

Apache Polaris

Polaris tracks the REST specification closely, served as the reference control in the September 2026 compatibility test [37], and is the strongest open default for a new multi-engine build. Its two-level role model (principal roles granted to service principals, catalog roles granted to principal roles) maps cleanly onto engine and agent identities, and each data privilege determines whether vended credentials are read-only or read-write [5][7]. Polaris 1.4 added separate encryption keys and IAM identities per catalog for multi-tenancy [35]. The limits sit at the edges: Hive Metastore federation requires a non-default build flag and supports one HMS with implicit authentication [9], and the project’s event stream needs a separate pipeline before it becomes an immutable audit archive [10]. Choose another option when your priority is governing several existing catalogs at once.

Project Nessie

Nessie governs catalog history, which no other catalog here does natively. Branch-level authorization means you can let a team commit freely on an experiment branch while only a pipeline principal can commit to main, using CEL (Common Expression Language) rules on the reference and the table path [12]. Nessie’s own documentation states that its authorization controls metadata only and that underlying data needs separate protection, and it still calls its Iceberg REST support experimental [11][12]. Nessie is the better fit when Git-style branching and cross-table commits matter more than broad federation.

Apache Gravitino

Gravitino is the catalog built for estates that already have several catalogs. A metalake is Gravitino’s top-level container that holds catalogs of different types (Hive, Iceberg, MySQL, PostgreSQL, Kafka) under one access model [15][18]. Its Iceberg REST service fronts Hive, JDBC, and other REST catalogs, and clients select a backend through the warehouse parameter [16]. Authorization pushdown translates a Gravitino grant into Apache Ranger policies so engines that read tables directly are also bound by it, a capability present since the 0.6 incubating releases [19]. Two details matter in deployment: access control on the Iceberg REST endpoint works only when it runs as an auxiliary service of the Gravitino server [16], and a native Glue catalog appears so far only in unreleased documentation [36]. Polaris is the better fit for a single greenfield catalog with the most complete REST coverage.

Unity Catalog

Databricks’ managed Unity Catalog has the deepest governance of the six when Databricks is your primary engine. Outside engines read, write, and create managed Iceberg tables through its REST endpoint, and catalog federation mounts Hive metastores, Glue, Snowflake, and several SaaS catalogs with Unity Catalog performing access checks and auditing [21][23]. The governance stops at the boundary in two places: federated external catalogs are read-only, and credential vending is unavailable for tables that carry row filters or column masks, so your most sensitive tables are the ones outside engines cannot reach through vending [22][23]. The open-source edition is a Linux Foundation AI & Data sandbox project under Apache 2.0, at release 0.5.1 [20][25]. Choose it when most writes run on Databricks.

AWS Glue Data Catalog

Glue is the natural catalog for an AWS-centered estate, because Lake Formation already holds the fine-grained permissions that AWS analytics engines enforce [40]. Lake Formation is AWS’s permission service for the Glue Data Catalog; it vends temporary credentials to third-party engines only after you enable that integration, either through IAM session tags or a full-table-access option [29]. The REST endpoint lacks view APIs, supports single-level namespaces, and rejects tables whose metadata exceeds 50 MB (5 MB per call for federated catalogs) [26][27]. Catalog federation for remote Iceberg REST catalogs has been generally available since November 24, 2025 [28]. It is the better fit when nearly all workloads run in AWS.

BigLake (Lakehouse Runtime Catalog)

Google renamed BigLake to Lakehouse on April 20, 2026, and the BigLake metastore REST catalog is now the Lakehouse runtime catalog, while API and IAM names still use “biglake” [34]. The catalog vends downscoped Cloud Storage tokens after checking IAM, and regional endpoints keep catalog requests in one location [31][32]. Its governance gaps are explicit: tables in this endpoint do not support row- or column-level security, table-level IAM is still in preview, and BigQuery DDL and DML cannot modify them [31][34]. Cross-cloud federation of Unity Catalog, Glue, Snowflake Horizon, and Workday catalogs arrived through 2026 in preview [34]. It fits best when BigQuery and Google Cloud are the center of the estate.

Lakekeeper is a seventh option worth a test if you want a lightweight self-hosted catalog: it is written in Rust, licensed under Apache 2.0, and uses OpenFGA for inherited permissions down to tables and views, with an OPA bridge for enforcement in Trino [38]. It remains pre-1.0, with v0.13.1 tagged June 30, 2026 [39].

Which Iceberg Catalog Has the Best Data Governance for Your Estate?

The catalog with the best governance is the one whose permission model covers every engine and environment that touches your tables, so the answer depends on the shape of the estate more than on any single feature. Broader platform choices sit in the guide to Apache Iceberg platforms.

If your estate is

Strongest governance fit

Trade-off you accept

Mostly Databricks, one cloud

Unity Catalog

Federated catalogs are read-only; masked tables are unavailable through vending

Mostly AWS engines

Glue with Lake Formation

No views over REST; AWS-only deployment

Mostly BigQuery

Lakehouse runtime catalog

No row- or column-level security on these tables today

New multi-engine build, any cloud or on-prem

Apache Polaris

Federation of existing catalogs is narrower

Several existing catalogs plus on-prem Hive

Apache Gravitino

Iceberg REST access control requires auxiliary-service mode

Branch-and-merge data workflows

Project Nessie

Iceberg REST support labeled experimental

Whichever you pick, one test exposes the real governance boundary: grant a principal read access to one table, then confirm that the same principal is denied a second table from every engine, every cloud, and every agent path that reaches the data.

What Is the Exit Path If You Start With a Managed Catalog?

A managed catalog keeps the exit open only if you design for it from the first table, because the files can move in an afternoon while grants, audit, and credential paths take months to rebuild. Catalog federation is the capability that makes a staged exit possible: a new catalog mounts the old one so engines can move before the tables do.

Use this sequence to keep the move a configuration change:

  1. Require every engine and agent to connect through the Iceberg REST API, never a vendor SDK path, so the endpoint URL is the only thing that changes [2].

  2. Keep table data and metadata in storage accounts you own and paths you control; note that the Lakehouse runtime catalog prohibits non-default write and metadata paths [31].

  3. Export grants as code from Unity Catalog, Lake Formation, or IAM before you need them, and map each to the target catalog’s role model.

  4. Stand up the target catalog and federate the source into it (Polaris, Gravitino, and Glue each mount remote REST catalogs) so readers switch first [8][16][28].

  5. Register existing tables in the target catalog by pointing at their current metadata file through the specification’s register endpoint [2].

  6. Move writers last and keep exactly one catalog accepting commits for each table, because two catalogs committing to one table fork its history.

  7. Rebuild audit and credential vending in the target, then rerun the deny test across every engine before you retire the source.

The ownership argument behind this sequence is in the CDO’s guide to a vendor-neutral data estate.

How NexusOne Approaches Iceberg Catalog Governance

NexusOne uses Apache Gravitino as the Iceberg catalog inside the NexusOne AI Data Control Plane, and adds the layer that makes one grant mean the same thing everywhere. Gravitino exposes the Iceberg REST protocol and federates Hive Metastore, Glue, other REST endpoints, and Databricks and Teradata metadata into one namespace, so Trino, Apache Spark, and Apache Kyuubi see the same tables whether they sit in an on-prem object store or a cloud bucket.

Identity arrives once through Keycloak, federated from Active Directory, Okta, LDAP, or SAML, and the same identity and policy model governs people, jobs, and AI agents. Policy is defined once in Apache Ranger and enforced across Trino, Spark, Kyuubi, and S3 object storage per object. Tagging a dataset in DataHub generates those policies at the bucket, schema, and table level, which is NexusOne’s answer to the access permissions and sensitive-data tags the EMA survey found most often reimplemented system by system. NexusOne’s view is that this tag-to-policy path is the governance step catalogs leave to you.

NexusOne also recommends two catalogs by purpose: internal catalogs for raw and refined data stay inside your estate, and shareable catalogs for data products are exposed so Databricks, Snowflake, or any Iceberg engine can read them under the same policy. The layer runs identically on-prem, in any cloud, hybrid, or air-gapped on a Kubernetes cluster you control, with 85+ integrated open-source components deployed under NexusOne’s published 5-5-5-5 model: about 5 minutes to provision, 5 hours to deploy, 5 days to the first governed workload, and 5 weeks to enterprise production, with Embedded Builders working alongside your team. A top-three US bank that had run three legacy Hadoop vendors connected 30 applications in under 4 weeks and eliminated more than $100 million in license and hardware costs.

To map your current catalogs onto one governed namespace, book an expert consultation.

Key Takeaways

  • The Iceberg REST specification standardizes how engines talk to a catalog, including credential vending and remote signing, but defines no grant or role model, so governance is each vendor’s design.

  • Polaris is the strongest open default for a new multi-engine build; Gravitino fits estates that must federate several existing catalogs; Nessie fits branch-and-merge workflows.

  • Unity Catalog, Glue, and the Lakehouse runtime catalog govern deeply inside their own cloud, and none runs in your data center.

  • Watch the boundary cases: read-only federated catalogs, no vending for masked tables in Unity Catalog, and no row- or column-level security on Lakehouse runtime catalog tables.

  • Keep the exit open by requiring the REST API from every client, owning storage paths, exporting grants as code, and moving writers last.

Frequently Asked Questions

Which Iceberg REST Catalog Has the Best Data Governance for Data Platforms?

It depends on where your engines run. Apache Polaris has the most complete open REST implementation with role-based access and credential vending, and suits new multi-engine builds. Apache Gravitino suits estates that need to federate Hive, JDBC, and other REST catalogs and push grants into Apache Ranger. Unity Catalog, Glue, and the Lakehouse runtime catalog govern best inside Databricks, AWS, and Google Cloud respectively.

Which Iceberg Data Catalog Delivers the Best Metadata Governance for Data Platforms?

For metadata governance across several systems, Apache Gravitino covers the most ground, because one metalake holds Hive, Iceberg, and relational catalogs under one role model with deny rules and audit logging. For version history of metadata, Project Nessie’s branch-level authorization is unique. Inside Databricks, Unity Catalog adds an audit system table and governed federation of Hive metastores and Glue, while open catalogs usually pair with companion tools such as DataHub.

What Is an Iceberg REST Catalog?

An Iceberg REST catalog is a catalog service that implements the HTTP API defined in the Apache Iceberg OpenAPI specification. Engines such as Spark, Trino, Flink, and DuckDB use it to list namespaces, load and commit tables, and request vended storage credentials, all without a vendor-specific client. Polaris, Nessie, Gravitino, Unity Catalog, Glue, and the Lakehouse runtime catalog all expose this interface.

Does the Iceberg REST Specification Define Access Control?

Only authentication. The specification defines an Authorization header, a 403 response for forbidden requests, and a header that lets clients accept vended credentials or remote signing. It defines no roles, privileges, or grant endpoints. Each catalog supplies its own model, from Polaris principal and catalog roles to Lake Formation permissions, so grants do not transfer when you change catalogs.

Can You Federate Unity Catalog, Glue, and Hive Metastore Into One Iceberg Catalog?

Partly, and the direction matters. Unity Catalog can mount Hive metastores and Glue as read-only foreign catalogs, Glue can mount remote Iceberg REST catalogs, and Google’s Lakehouse runtime catalog can mount Unity Catalog, Glue, and Snowflake Horizon in preview. Apache Gravitino and Apache Polaris federate Hive and other REST catalogs while running on infrastructure you control, including on-prem.

References

  1. More Tools, Less Control: The 2026 State of the Data Estate Report, NexusOne, published September 30, 2026 (survey conducted by Enterprise Management Associates (EMA), September 2026, n=202 data and analytics decision-makers at organizations with 500+ employees in the US, Canada, the UK, and France). https://nexus0ne.substack.com/p/more-tools-less-control-the-2026 (retrieved October 20, 2026)

  2. Apache Iceberg REST Catalog API (OpenAPI specification), Apache Iceberg. https://github.com/apache/iceberg/blob/main/open-api/rest-catalog-open-api.yaml (retrieved October 20, 2026)

  3. Apache Polaris Graduates to Top-Level Project, Apache Polaris Blog, February 19, 2026. https://polaris.apache.org/blog/2026/02/19/apache-polaris-graduates-to-top-level-project/ (retrieved October 20, 2026)

  4. Releases, apache/polaris, GitHub (Apache Polaris 1.8.0, September 28, 2026). https://github.com/apache/polaris/releases (retrieved October 20, 2026)

  5. Access Control, Apache Polaris 1.7.0 Documentation. https://polaris.apache.org/releases/1.7.0/managing-security/access-control/ (retrieved October 20, 2026)

  6. Open Policy Agent (OPA) Integration, Apache Polaris 1.7.0 Documentation. https://polaris.apache.org/releases/1.7.0/managing-security/external-pdp/opa/ (retrieved October 20, 2026)

  7. Overview, Apache Polaris Documentation. https://polaris.apache.org/docs/overview (retrieved October 20, 2026)

  8. Apache Polaris Documentation, latest release (Federation: Iceberg REST, Hive Metastore, BigQuery Metastore). https://polaris.apache.org/releases/latest/ (retrieved October 20, 2026)

  9. Hive Metastore Federation, Apache Polaris 1.7.0 Documentation. https://polaris.apache.org/releases/1.7.0/federation/hive-metastore-federation/ (retrieved October 20, 2026)

  10. Apache Polaris Events and Audit Logging: A Production Guide, Dremio, September 25, 2026. https://www.dremio.com/blog/apache-polaris-events-and-audit-logging-a-production-guide/ (retrieved October 20, 2026)

  11. Nessie and Iceberg REST, Project Nessie Documentation. https://projectnessie.org/guides/iceberg-rest/ (retrieved October 20, 2026)

  12. Authorization (Access Control), Project Nessie Documentation. https://projectnessie.org/nessie-latest/authorization (retrieved October 20, 2026)

  13. Releases, projectnessie/nessie, GitHub. https://github.com/projectnessie/nessie/releases (retrieved October 20, 2026)

  14. projectnessie/nessie repository (Apache-2.0 license), GitHub. https://github.com/projectnessie/nessie (retrieved October 20, 2026)

  15. The Apache Software Foundation Announces Two New Top-Level Projects, The Apache Software Foundation, June 3, 2025. https://news.apache.org/foundation/entry/the-apache-software-foundation-announces-two-new-top-level-projects (retrieved October 20, 2026)

  16. Iceberg REST Catalog Service, Apache Gravitino 1.3.1 Documentation. https://gravitino.apache.org/docs/1.3.1/iceberg-rest-service (retrieved October 20, 2026)

  17. Gravitino 1.3.0 Release Notes, Apache Gravitino Blog, June 29, 2026. https://gravitino.apache.org/blog/gravitino-1-3-0-release-notes (retrieved October 20, 2026)

  18. Access Control, Apache Gravitino 1.2.0 Documentation. https://gravitino.apache.org/docs/1.2.0/security/access-control (retrieved October 20, 2026)

  19. Authorization Push-down, Apache Gravitino 0.6.0-incubating Documentation. https://gravitino.apache.org/docs/0.6.0-incubating/security/authorization-push-down (retrieved October 20, 2026)

  20. unitycatalog/unitycatalog repository README, GitHub. https://github.com/unitycatalog/unitycatalog (retrieved October 20, 2026)

  21. External Access to Databricks Data, Databricks Documentation, updated October 14, 2026. https://docs.databricks.com/aws/en/external-access/ (retrieved October 20, 2026)

  22. Unity Catalog Credential Vending for External System Access, Databricks Documentation. https://docs.databricks.com/aws/external-access/credential-vending (retrieved October 20, 2026)

  23. What Is Catalog Federation?, Databricks Documentation. https://docs.databricks.com/aws/en/query-federation/catalog-federation (retrieved October 20, 2026)

  24. Audit Log System Table Reference, Databricks Documentation. https://docs.databricks.com/en/admin/system-tables/audit-logs.html (retrieved October 20, 2026)

  25. Releases, unitycatalog/unitycatalog, GitHub (Unity Catalog 0.5.1). https://github.com/unitycatalog/unitycatalog/releases (retrieved October 20, 2026)

  26. Connecting to the Data Catalog Using AWS Glue Iceberg REST Endpoint, AWS Glue Developer Guide. https://docs.aws.amazon.com/glue/latest/dg/connect-glu-iceberg-rest.html (retrieved October 20, 2026)

  27. Considerations and Limitations When Using AWS Glue Iceberg REST Catalog APIs, AWS Glue Developer Guide. https://docs.aws.amazon.com/glue/latest/dg/limitation-glue-iceberg-rest-api.html (retrieved October 20, 2026)

  28. AWS Glue Announces Catalog Federation for Remote Apache Iceberg Catalogs, AWS, November 24, 2025. https://aws.amazon.com/about-aws/whats-new/2025/11/aws-glue-catalog-federation-remote-apache-iceberg-catalogs (retrieved October 20, 2026)

  29. Using Lake Formation Application Integration, AWS Lake Formation Developer Guide. https://docs.aws.amazon.com/lake-formation/latest/dg/using-cred-vending.html (retrieved October 20, 2026)

  30. Logging AWS Glue API Calls With AWS CloudTrail, AWS Glue Developer Guide. https://docs.aws.amazon.com/glue/latest/dg/monitor-cloudtrail.html (retrieved October 20, 2026)

  31. Use the BigLake Metastore Iceberg REST Catalog, Google Cloud Documentation. https://docs.cloud.google.com/biglake/docs/blms-rest-catalog (retrieved October 20, 2026)

  32. Credential Vending, Google Cloud Lakehouse Documentation, updated October 15, 2026. https://docs.cloud.google.com/biglake/docs/credential-vending?hl=en (retrieved October 20, 2026)

  33. Iceberg REST Catalog Now Supported in BigLake Metastore (GA), Google Cloud Blog, November 19, 2025. https://cloud.google.com/blog/products/data-analytics/biglake-metastore-now-supports-iceberg-rest-catalog/ (retrieved October 20, 2026)

  34. Lakehouse for Apache Iceberg Release Notes, Google Cloud Documentation. https://docs.cloud.google.com/lakehouse/docs/release-notes (retrieved October 20, 2026)

  35. What’s New in Apache Polaris 1.4: Advanced Multi-Tenancy and Streamlined Deployment, Snowflake Engineering Blog, April 23, 2026. https://www.snowflake.com/en/engineering-blog/apache-polaris-1-4-release/ (retrieved October 20, 2026)

  36. AWS Glue Catalog, Apache Gravitino Documentation (unreleased “next” version). https://gravitino.apache.org/docs/next/aws-glue-catalog (retrieved October 20, 2026)

  37. The 2026 Iceberg REST Catalog Compatibility Report, Alex Merced, Data Lakehouse Hub, September 10, 2026. https://datalakehousehub.com/blog/iceberg-rest-catalog-compatibility-report-2026 (retrieved October 20, 2026)

  38. Lakekeeper Documentation. https://docs.lakekeeper.io (retrieved October 20, 2026)

  39. Releases, lakekeeper/lakekeeper, GitHub (v0.13.1, June 30, 2026). https://github.com/lakekeeper/lakekeeper/releases (retrieved October 20, 2026)

  40. Read and Write S3 Iceberg Table Using AWS Glue Iceberg REST Catalog From Open Source Apache Spark, AWS Big Data Blog, December 4, 2024. https://aws.amazon.com/blogs/big-data/read-and-write-s3-iceberg-table-using-aws-glue-iceberg-rest-catalog-from-open-source-apache-spark (retrieved October 20, 2026)

ABOUT

1115 Howell Mill Rd
Suite 430,
Atlanta, GA 30318
An Insight Partners Company


Product Updates and News

@2026 NexusOne® - All rights reserved.

ABOUT

1115 Howell Mill Rd
Suite 430,
Atlanta, GA 30318
An Insight Partners Company


Product Updates and News

@2026 NexusOne® - All rights reserved.

ABOUT

1115 Howell Mill Rd
Suite 430,
Atlanta, GA 30318
An Insight Partners Company


Product Updates and News

@2026 NexusOne® - All rights reserved.