DuckDB-Iceberg v1.5.3, PostgreSQL Hash Join Tuning, & Relational Algebra to SQL
Today's Highlights
Today's highlights include new Iceberg features in DuckDB v1.5.3, enhancing data lake capabilities with MERGE INTO and ALTER TABLE. We also explore PostgreSQL performance tuning with the enable_hashjoin GUC and a unique tool for learning relational algebra that compiles to SQL.
New DuckDB-Iceberg Features in v1.5.3 (DuckDB Blog)
Source: https://duckdb.org/2026/05/29/new-iceberg-features.html
DuckDB's latest v1.5.3 release introduces significant enhancements for working with Apache Iceberg tables, crucial for data lake architectures. Key new features include the MERGE INTO statement, enabling efficient upsert and update operations on Iceberg tables directly from DuckDB. Additionally, ALTER TABLE commands now support a broader range of schema modifications, providing more flexibility for evolving data models within your data lake.
The update also brings comprehensive support for Iceberg REST Catalogs, simplifying metadata management and integration with existing Iceberg environments. Furthermore, improved handling of partition transforms and full support for Iceberg V3 features ensure that DuckDB remains a powerful, performant, and embedded solution for analytical workloads on modern data lake formats. These additions empower data engineers to build more robust and dynamic data pipelines using DuckDB as an in-process query engine.
Comment: These Iceberg updates solidify DuckDB's role in the data lake ecosystem, making complex data pipeline operations like upserts much more straightforward for analytics.
All Your GUCs in a Row: enable_hashjoin (Planet PostgreSQL)
Source: https://postgr.es/p/9nv
Christophe Pettus explores the enable_hashjoin GUC (Grand Unified Configuration) parameter in PostgreSQL, offering a practical guide for diagnosing query performance issues. Hash joins are often efficient, but can lead to 'spilling' to disk when memory is insufficient, severely impacting performance. By temporarily disabling hash joins with enable_hashjoin = off, developers can force the planner to use alternative join methods, such as nested loop or merge joins.
This technique helps to isolate whether a hash join is indeed the bottleneck in a complex query. If a query performs better with hash joins disabled, it points to a need for tuning memory parameters (like work_mem) or re-evaluating the query structure. Understanding and manipulating GUCs like this is a fundamental skill for PostgreSQL performance tuning, providing direct control over the query optimizer's behavior for diagnostic purposes.
Comment: This is a great, actionable tip for PostgreSQL users facing slow queries. Disabling enable_hashjoin is a quick way to pinpoint if your hash joins are spilling and need tuning.
coddpiece: Watch Relational Algebra Become SQL (Planet PostgreSQL)
Source: https://postgr.es/p/9nr
Christophe Pettus introduces 'coddpiece', a fascinating project aimed at helping developers understand the foundational link between relational algebra and SQL. The tool allows users to construct expressions based on relational algebra concepts (like projection, selection, union, join) and then observes how these theoretical constructs compile into actual SQL queries. This bridge from theory to practice is invaluable for anyone looking to deepen their understanding of how databases operate and how to write more efficient and semantically correct SQL.
By interactively seeing relational algebra transform into SQL, users can gain a clearer intuition for query optimization principles and the underlying logic behind database operations. It serves as an excellent educational resource for both beginners and experienced developers seeking to solidify their knowledge of database fundamentals. This interactive approach demystifies complex SQL constructs by revealing their mathematical origins.
Comment: As someone who's taught SQL, a tool like coddpiece is a game-changer for grasping relational algebra. It makes abstract concepts concrete and improves query writing by building foundational understanding.
Top comments (0)