http://starrocks.io logo
Join Slack
Powered by
<@U0A71G31CDV> in which version is starrocks expec...
# questions-and-troubleshooting
e
@Rocky in which version is starrocks expected to support querying iceberg variant type with the shredding performance ?
r
StarRocks supports querying the Apache Iceberg VARIANT data type starting from v4.1. However, for the specific "shredding" performance optimization you are looking for, the status is as follows: * Current Status (v4.1): Support for the 
VARIANT
type in Iceberg (Parquet format) was introduced in version 4.1, but it is currently limited to the unshredded form. This means that while you can query the data, it is typically processed as a binary blob rather than leveraging the columnar "shredded" sub-fields for maximum performance. * Roadmap (v4.2 and beyond): High-performance shredding (the ability to extract frequently used sub-fields into independent columns for sub-second query performance) is a major focus of the StarRocks 2026 Roadmap. According to community roadmap updates and recent technical blogs, advanced 
VARIANT
support and full Iceberg V3 specification compatibility (which includes shredding) are slated for the v4.x cycle throughout 2026, with v4.2 already expanding 
VARIANT
support to other catalogs like Paimon. Summary of Support Feature | StarRocks Version Basic Iceberg VARIANT (Unshredded) | Supported in v4.1+ Iceberg VARIANT Shredding Performance | On the 2026 Roadmap (expected in v4.2 or later 4.x releases) For real-time updates on this feature, you can follow the development in StarRocks GitHub Issue #55949 or the Roadmap 2026. References • data_source/catalog/iceberg/iceberg_catalog.md • sql-reference/data-types/semi_structured/VARIANT.md • data_source/catalog/iceberg/procedures.md • data_source/feature-support-data-lake-analytics.md
StarRocks began supporting the VARIANT data type for Apache Iceberg tables in version 4.1 (released in September 2026). However, there are specific details regarding its current performance and support for shredding: * Current Support (v4.1): As of version 4.1, StarRocks supports querying the 
VARIANT
type for Iceberg tables in Parquet format, but it currently only supports unshredded variant values. This means the engine can read the binary-encoded variant blobs but cannot yet take full advantage of the "shredding" optimization where sub-fields are stored as separate columnar data. * Performance Expectation: Shredding is required to achieve high-performance columnar scanning and predicate pushdown for semi-structured data. StarRocks has officially stated a goal to achieve full support for Apache Iceberg v3 table formats (which includes the Variant type with shredding) within 2026. * Roadmap: Internal roadmaps and community updates (such as those from the Cisco Webex case study) indicate that advancing Variant Shredding is a top priority to close the performance gap with other specialized semi-structured data engines. While a specific version number for shredding support has not been formally announced beyond v4.1, it is expected to land in a subsequent minor release (likely v4.2 or later) in late 2026. Summary of Version Support Feature | StarRocks Version Iceberg VARIANT Type (Unshredded) | v4.1 Iceberg VARIANT Shredding Performance | Expected later in 2026 (v4.2+) If you need the highest possible performance for semi-structured data today, the official recommendation is to continue using the Flat JSON optimization (using 
JSON
type with generated columns or specialized indexing) until the native Iceberg shredding support is finalized.
2 Views
Open in Slack
PreviousNext
Powered by