What Actually Happens When Two Engines Write the Same Iceberg Table at Once?
Somewhere in your platform, right now, a Spark job and a streaming writer are heading toward the same Apache Iceberg table, and they will…
The archive
Writing, page 4
593 posts on Apache Iceberg, lakehouse architecture, data engineering and applied AI.
Variant Shredding Explained: How Iceberg Gets Columnar Performance From Messy JSON
The Variant type in Apache Iceberg v3 gets described in one sentence so often that the sentence has started doing damage: "store JSON…
Who Actually Owns an Iceberg Table? Managed, External, and the New Vocabulary of Lakehouse Control
Two architects are arguing in a design review about whether to use "managed" Iceberg tables, and the argument is unresolvable, because they…
Mastering Apache Iceberg v3 Deletion Vectors for High-Throughput Streaming Ingest
Here is a bill that surprises teams every quarter. A Flink pipeline streams change data capture events into an Apache Iceberg table, a few…
The Decoupled Data Lakehouse: Multi-Engine Freedom with Open REST Catalogs
Every few years a data team discovers, mid-contract-renewal, exactly how much of their platform they do not control. The data sits in the…
The Five Layers of an Agentic Lakehouse
Every architecture era gets its reference diagram. The warehouse era had its star schemas and its staging-to-mart flow. The big data era…
Goal-Directed Data Quality Agents: Anomaly Quarantine on Apache Iceberg
Every data platform has a quality system, and most of them are the same system: a few hundred rules, written after incidents, checking the…
Managing the TCO of Agentic Analytics: Token Budgets, Query Throttles, and the Economics of Autonomy
The AI budget conversation changed its tone this year, and the numbers explain why. Industry surveys of enterprise AI spending in 2026 keep…
Metric Contracts in 2026: Standardizing Business Logic Across Multi-Agent Frameworks
For twenty years, the cost of an ambiguous metric was a meeting. Two dashboards disagreed, two teams defended their numbers, someone…
Policy-Aware Lakehouse Telemetry: Building Auditable AI Records on Apache Iceberg for the EU AI Act Era
Every article in my agentic lakehouse series has quietly assumed a plane that sees everything: the catalog's audit stream, the semantic…