dataaaaa!

Amazon Redshift patch 202 is currently being deployed across all supported AWS regions. This update follows a phased rollout strategy that typically takes several weeks to complete. Data engineers should monitor cluster maintenance windows for version availability.

Snowflake has renamed the dynamic tables immutability constraint to frozen regions. The IMMUTABLE WHERE clause is now FROZEN WHERE, though the change is backward compatible. This update simplifies terminology for data engineers managing stateful transformations and backfill logic.

Snowflake introduced a Sensitive Data Access report in public preview to track user access to classified tables. It generates a view mapping user-table-role combinations from query history. The report creates a view that lists user-table-role combinations identified in query and access history.

Snowflake Cortex AI now supports Gemini 3.5 Flash for multimodal video and audio analysis. This allows engineers to extract sentiment and behavioral insights directly within Snowflake.

Polars 1.41 optimizes Parquet metadata decoding by replacing generic Thrift decoders with hand-written ones, significantly speeding up scans for wide tables. The previous decoder was auto-generated from the Thrift IDL definitions, which are general-purpose and built to describe arbitrary schemas.

Snowflake introduced a public preview for managing data protection policies directly within the Snowsight interface. Engineers can now create data protection policies from the Dashboard and monitor protected objects via a dedicated tab, streamlining governance workflows without manual SQL.

Snowflake now allows users to build and register custom container images for Notebooks and ML Jobs. This feature enables engineers to include specific packages, meet compliance and security requirements, and integrate with existing CI/CD pipelines while ensuring environment reproducibility.

Google has released an updated version of the Simba ODBC driver for BigQuery. This update ensures improved connectivity and performance for external applications interfacing with the data warehouse. An updated version of the Simba ODBC driver for BigQuery is now available.

Snowflake Data Clean Rooms now features the DCR Agent in public preview, a conversational interface for joining collaborations and running audience overlap analyses.

Snowflake introduced CUSTOM_INCREMENTAL refresh mode for dynamic tables, allowing engineers to define manual MERGE or INSERT logic. This provides control over complex patterns like stream-static joins and stateful aggregation while Snowflake handles scheduling and transactional guarantees.

Snowflake introduced an adaptive refresh mode for dynamic tables that optimizes performance by switching between incremental and full refreshes. ADAPTIVE refresh uses incremental refresh by default but automatically reinitializes when heuristics detect that a full refresh is more efficient.

The Data Science Agent for Colab Enterprise and BigQuery is now generally available. This tool streamlines data engineering workflows by assisting with analysis and automation. The Data Science Agent (DSA) for Colab Enterprise and BigQuery is now generally available (GA).

Orchestra released State-Aware Orchestration (SAO) for dbt Core, a managed Python package that tracks model state to skip redundant runs. It uses source freshness checks to save costs and speed up pipelines. Models are skipped if there is no new data and no code has changed.

Flink Kubernetes Operator 1.15.0 introduces native Kubernetes Conditions for machine-readable status signals and adds Logback support. It ensures Flink 2.2 compatibility and bundles the Dropwizard metric reporter. The operator now exposes a standard Kubernetes Condition in the status field.

The Openflow Connector for Oracle now supports replicating data from multiple Pluggable Databases using a single instance. This update eliminates the need for separate connectors per PDB. Replicating tables from multiple PDBs required a separate connector instance for each PDB.

The pg_infer extension for PostgreSQL 18+ treats transformer models as first-class data sources. It exposes gate activations and embeddings as SQL-queryable relations. The model becomes a first-class data source -- a set of relations the planner can scan, filter, and join. It runs on CPUs.

DuckDB v1.5.3 introduces the Quack remote protocol as a core extension, enabling client-server functionality. It adds Iceberg support for MERGE INTO and partitioned table updates.

Snowflake Cortex AI_EXTRACT now provides confidence scores between 0 and 1 to quantify model certainty. This allows engineers to build human-in-the-loop workflows by flagging low-scoring extractions. Use scores to set thresholds and flag low-scoring extractions for human review.

Snowflake now supports JWT-based authentication for AWS Workload Identity Federation via the Python connector. This method uses the GetWebIdentityToken API to produce a standards-based JWT. Snowflake recommends upgrading to the JWT-based method by setting the outbound token environment variable.

Snowflake overhauled its dynamic tables documentation with 30 new pages covering the pipeline lifecycle. Key additions include SCD Type 1 deduplication patterns and migration guides from streams and tasks. Every page includes complete, copy-pasteable SQL examples with expected output.

Snowflake now supports attaching storage lifecycle policies to dynamic tables to automate row deletion or archiving. Policies run asynchronously from table refreshes, treating expired data as a frozen region. The policy runs asynchronously on its own schedule to delete or archive those rows.

Snowflake Data Clean Rooms version 15.6 introduces performance improvements and updates to private preview features. Engineers can verify their environment version by running SELECT FROM SAMOOHABYSNOWFLAKELOCALDB.ADMIN.VERSION;* to ensure the latest API updates are active.

Snowflake Cortex AI Function Studio simplifies building production-ready functions for multimodal data. It automates prompt engineering and model selection to optimize quality. AI Function Studio automates prompt engineering, model selection, benchmarking, and optimization for complex workflows.

Snowflake now supports general availability for writing to managed Apache Iceberg v2 and v3 tables via external engines like Apache Spark. This utilizes the Iceberg REST protocol through Snowflake Horizon, allowing users to read and write to Snowflake-managed Iceberg v2 and v3 tables.

BigQuery now re-executes queries to proactively detect performance or functional regressions. These background runs have no side effects and incur no costs. Data access logs may show bigquery-adminbot@system.gserviceaccount.com when BigQuery re-executes an instruction.