Search

A Tale of Two Houses: A Comparison Between Streamhouse and Lakehouse

The document compares Streamhouse and Lakehouse architectures, defining Streamhouse as a real-time, current-state data layer that propagates changes to consumers, while Lakehouse stores file-based data with a metadata layer for batch-processed analytics. It explains their distinct properties, failure modes, and how they can complement each other in a hybrid data architecture, including practical use cases and terminology like Lightning Tables and open Iceberg/Delta formats.

GitHub - bhbyuh/lakehouse-Implementation · GitHub

A self-contained Docker-based lakehouse playground integrating Spark, Iceberg, Hive Metastore, Trino, Kafka, MinIO, Airflow, and Superset for end-to-end data ingestion, streaming, maintenance, orchestration, and BI; includes setup steps, architecture diagram, project layout, and runtime URLs.

Getting Started with DuckLake: Build a Lakehouse on DuckDB in Under an Hour - KDnuggets

This tutorial sets up a local DuckLake lakehouse with Python and DuckDB, using Parquet files for data and a SQL database for metadata. It begins by generating sample order data and creating the lake; later steps are outlined for data changes, time travel, and maintenance.

The semantic layer turns lakehouses into AI decision backbones - SiliconANGLE

Oracle reframes lakehouses as AI decision backbones, introducing Autonomous AI Lakehouse with a shared semantic layer that queries data in place across multiple sources and emphasizes real-time access for AI agents.

Direct Iceberg access in AlloyDB | Google Cloud Blog

AlloyDB announces direct Iceberg access in preview, GA for AlloyDB sync, and GA for Lakehouse federation, enabling unified queries across live operational data and lakehouse analytics. The borderless Lakehouse keeps a single data copy accessible to multiple engines, with real-world gains (e.g., Etsy) in query speed.

Databricks aims data lakehouse at media and entertainment - Bytes Europe

Databricks releases a media-focused data lakehouse, leveraging Apache Spark to democratize access to unstructured video data for Sega and others, enabling faster analytics, improved game balancing, and accelerated use of data for AI, BI, and personalization.

Run dbt on Snowflake and DuckDB with shared Iceberg tables | Greybeam

This guide demonstrates a multi-engine dbt project where Snowflake and DuckDB collaboratively build and access the same Iceberg tables via the Horizon catalog, replacing prior bucket/IAM setups with Snowflake-managed storage and per-table credentials. It outlines base, intermediate, and marts layers, test practices, and required releases and permissions.

Apache Iceberg on Snowflake got a lot simpler in 2026, and dbt's new catalog config is what makes it click | dbt Labs

dbt Summit 2026 showcases simplified use of Apache Iceberg on Snowflake, driven by Snowflake’s Horizon catalog and a new dbt catalog config. The changes let teams run governed Iceberg tables with minimal infrastructure, enabling zero-Snowflake compute through DuckDB reads and built-in governance tooling, while DuckLake remains an alternative.

Amazon S3 Tables で Apache Hive テーブルから Apache Iceberg へのマネージドな移行が可能に | Amazon Web Services ブログ

AWS S3 Tables now offers a managed migration from Hive to Iceberg for Parquet tables in Glue, enabling in-place, low-cost upgrades with automated validation, compatibility checks, and minimal downtime. The feature supports integration with AWS tools and rest APIs, and is available in multiple regions.

A data lake alone doesn’t make an AI strategy | Computerwoche

Translated title. Original: Ein Data Lake allein macht noch keine KI-Strategie | Computerwoche

Large companies often rely on costly, fragile data pipelines, and federated data systems bring added governance and synchronization demands. Making these systems ready for AI agents is difficult because companies must improve them while keeping existing operations running.