Today, we are excited to announce Starburst Enterprise 476-e LTS. This release pushes forward our commitment to multi-cluster data architecture, deepens our Iceberg leadership, and advances AI readiness while delivering steady gains in performance and usability.
Collectively, 476-e introduces significant improvements in the following areas:
- Introducing the Starburst Portal, coupling the Starburst Data Catalog (Public Preview) and the Starburst Gateway (Private Preview)
- Data Product Sharing (Private Preview)
- Iceberg Data Products
- Enhanced Starburst AI Agent capabilities (Private Preview)
- Iceberg v3 and Iceberg branching
- Significant performance gains
Starburst Portal helps you manage your multi-cluster deployments
Scalable by design, the future of the Starburst data platform is multi-cluster. With the release of Starburst Portal, we take a major step toward enabling that future for our customers who desire – or require – a multi-cluster environment. By combining the Starburst Data Catalog and Starburst Gateway into a single entry point, the Starburst Portal provides the foundation for seamless cluster management, metadata control, and connectivity. The Starburst Portal will act as the foundation for building a natively multi-cluster platform, separating services from compute to better serve customer multi-cluster needs and simplify large-scale deployments.
The portal is composed of two major components:
- Starburst Data Catalog (Public Preview)
- Starburst Gateway (Private Preview)
Starburst Data Catalog
The Starburst Data Catalog is a self-hosted, AWS Glue-compatible metastore designed to replace the legacy Hive Metastore with a more secure, scalable, and interoperable foundation. It is deployable in Kubernetes environments using Helm. Importantly, it allows you to run self-managed SEP clusters with Glue-compatible metadata management without requiring access to AWS infrastructure, offering an alternative enterprise-grade catalog designed for Trino and beyond.
The Starburst Data Catalog serves as the initial core component enabling our multi-cluster architecture. With a single control plane for technical metadata, you get faster queries, cleaner governance, and better collaboration across teams.
Starburst Gateway
The Starburst Gateway is a single entry point for Starburst Enterprise clusters. This load balancing solution routes traffic to Starburst clusters based on a set of configurable rules, allowing you to manage complex deployments across multiple clusters.
In this sense, it acts as the second main foundational component for our multi-cluster vision, abstracting away complexities normally associated with multi-cluster management for a variety of critical enterprise use cases.
Data products power on-premises and hybrid deployments
Data products in 476-e now include an extended suite of capabilities supporting both our core Icehouse and multi-cluster vision.
Data Product Sharing (Private Preview)
Data products sharing allows you to replicate and share governed datasets across clusters. This means that rather than recreating the same data product in multiple environments, teams can distribute an identical, access-controlled version that is ready for use.
This added capability promotes reusability, safeguards data integrity, and supports the strict infrastructure boundaries common in high-compliance, on-premises, and hybrid deployments. Interested in seeing data product sharing and how it works with our Starburst Agent? Check out this video to see the two features in action.
Iceberg data products (Generally Available)
Iceberg Data Products simplify reuse by packaging data with governance and metadata built in. Teams can now share consistent, trusted datasets in the Iceberg table format, reducing duplication and accelerating collaboration.
Improved user experience around Data Products and BIAC
The 476-e release introduces streamlined interfaces for managing Data Products and BIAC. These updates make discovery, configuration, and monitoring easier, helping data teams focus on insights rather than setup.
Automatically enrich data product metadata (Private Preview)
We’re introducing AI Enrichment for data products, a capability that applies machine learning models to automatically generate column-level descriptions, dataset summaries, product overviews, and tags directly from schema and contextual information.
This eliminates the hardest part of a data product – documenting it. To view an example of this feature in action, take a look at our video demonstration.
Starburst AI Agent (Private Preview)
The 476-e release builds on the AI momentum started in 474-e, specifically regarding the Starburst AI agent. This AI Agent is designed to make your data more discoverable, understandable, and actionable, helping organizations translate raw data into real business outcomes.
Generate insights with natural language
Leveraging data products, the Agent enables users from across your organization to interact with data using natural language. It creates queries, runs analyses, and tailors responses based on user roles, accelerating insights without requiring deep SQL expertise.
Collaborate across clusters
When combining the power of the agent with data product sharing, insights generated by the Agent can flow across clusters and teams. This ensures that curated, governed data products are not only accessible but reusable, creating a foundation for organization-wide collaboration and AI adoption.
Starburst takes the next step with Apache Iceberg
Apache Iceberg has become the standard for building modern data lakehouses, and with 476-e we’ve deepened our commitment to the format. By supporting the latest Iceberg features, Starburst ensures that Iceberg users can take advantage of faster performance, richer data types, and safer development workflows, all within the same trusted platform.
Support for Iceberg v3
476-e introduces full support for Iceberg v3, a major leap forward for the format. Here, users benefit from binary deletion vectors, which speed up updates and deletes. Iceberg v3 also extends support for richer data types, such as variant data for semi-structured data, nanosecond-precision timestamps, and row-level lineage for stronger governance. Collectively, these enhancements mean faster queries and improved auditability across the board.
Support for Iceberg branching
We’re also introducing support for Iceberg branching, giving data teams Git-style version control for their tables. Iceberg branching makes it possible to isolate changes, run backfill jobs, or experiment with transformations without risking production data. It also enables safer collaboration across teams by allowing multiple workflows to run in parallel. Together with Starburst, branching helps organizations move faster while maintaining trust in their data.
Starburst Enterprise performance improvements
Performance is always top of mind, and this release is no different. Starburst Enterprise 476-e LTS returns consistently faster queries, especially for queries that include aggregations and joins. TPC-DS testing shows Starburst Enterprise LTS 476 consumed 30% less CPU time and improved query speed by 20% when compared to Starburst Enterprise 468 LTS.
Note: Benchmark performed August 2025 using TPC-DS sf1000, with a concurrency of 5.
Starburst ODBC Driver
Starburst has enhanced our ODBC support with an improved V3 ODBC driver, delivering over 400% performance improvements in testing. It is a drop-in replacement for version 2 and is currently available for Windows, particularly for Tableau or Power BI users.
If you are using another BI tool, this is a great opportunity to start evaluating whether ODBC V3 might work for you to improve compatibility and performance.
Critical and breaking changes
This Starburst Enterprise release includes the following critical and breaking changes, summarized here. Please view the full documentation for more information, as well as the complete list of breaking changes.
Starburst Portal Deployment
As of version 476-e, Starburst Gateway and Starburst data catalog are now deployed together using a unified Helm chart called starburst-portal. This new deployment artifact simplifies installation and ensures compatibility with the latest features and fixes.
Note: The previous individual Helm charts for starburst-gateway and starburst-catalog are no longer supported. For more information, please review the latest documentation for important configuration updates to Starburst Gateway and Starburst Catalog.
JVM Memory configuration
The –sun-misc-unsafe-memory-access=allow JVM option is required when running Trino with the BigQuery or Snowflake connectors. Add this option to your Trino JVM configuration to ensure proper connector operation.
Starburst Admin Deployment Update
While distinct from the LTS, we want to remind you about a previous breaking change to the Starburst Admin deployment option. The Starburst Admin 1.14.0 release is now mandatory for supported Starburst Enterprise versions due to required updates for Java 24 compatibility. Testing has been removed for Starburst Enterprise 443-e.
For more information, please consult the Starburst Admin release notes as needed.
AWS SDK V1
The AWS SDK v1 will reach end of support on December 31, 2025. In our previous May LTS, we updated the default configuration for Hive, Delta Lake, and Hudi to be V2. However, users still have the option to utilize V1 by configuring the hive.metastore property to hive.metastore=glue-v1.
In this 476 LTS, we added support for AWS Glue v2 to Iceberg. You can enable it using the iceberg.catalog.type=glue_v2 configuration property.
We are asking customers to migrate to utilize Glue V2 as soon as possible. This will help mitigate disruption for your workloads in the future. We are expecting to deprecate this V1 adoption before end of life support. Additionally, we are expecting to deprecate this V1 adoption before end of life support is reached.
Configuration changes
Several configuration changes have also been implemented in this release. The changes include:
- The query.max-written-data-size query management property has been renamed to query.max-write-physical-size.
- The hive.s3.storage-class-filter configuration property has been renamed to hive.s3-glacier-filter in the Hive connector.
- The hive.file-status-cache-tables.excluded configuration property has been renamed to hive.file-status-cache.excluded-tables.
- The hive.s3.storage-class-filter configuration property has been renamed to hive.s3-glacier-filter.
- The kudu.schema-emulation.enabled property has been removed and replaced with the new property kudu.schema-emulation.type.
- The kafka-event-listener.client-config-overrides configuration property has been removed.
Please view the full documentation for more information.
Upgrade to the August 476 LTS Today
The 476-e LTS release marks a significant milestone in enabling organizations to scale and govern their data platforms.
