Latest Articles: page 8
692 posts from Alex's blogs and newsletters, newest first. Thoughts on tech, data, policy, and philosophy.
Apache Fluss and Kafka Solve Different Problems in an Iceberg Pipeline
Fluss puts a columnar, indexed hot tier between Kafka and Iceberg. Here's what it changes structurally, what Kafka still does better, and how to bench...
Serving Sub-Second Queries Over an Iceberg Lakehouse With a Hot Tier
A lakehouse cannot serve sub-second queries over seconds-old data. A hot tier in front solves it, with consequences for consistency, governance....
Surviving Commit Conflicts When Dozens of Writers Hit the Same Iceberg Table
Commit conflicts multiply with writer count, and AI agents introduce unpredictable write patterns....
The Jackson 3 Problem in Apache Iceberg, and What It Means for Your Code
Jackson 3 changes everything: package names, unchecked exceptions, flipped defaults. Here's what breaks, why the engines are fine and your service isn...
Wiring an AI Agent to Apache Polaris with the Model Context Protocol
The catalog is the right attachment point for AI agents working against a lakehouse. Here's how to wire the official Polaris MCP Server and add the re...
Governing Iceberg Tables Across Regions Without Three Sets of Permissions
Catalog federation gives you one authorization model and one audit point across regions. Here's what it solves, what it doesn't, and how to build....
Federating Oracle With an Open Lakehouse Instead of Migrating It
Federate first so analytics work now, migrate what benefits from migrating, and leave the rest where it is indefinitely....
The Parquet Versioning Problem, and Why Iceberg Cares About It
Parquet files have a version field that doesn't reliably signal feature requirements. A new versioning discipline is coming, borrowing from Iceberg's....
Building Iceberg Pipelines in Python Without Standing Up Spark
A large share of production transformations fit comfortably on one machine. PyIceberg, DuckDB, and branch isolation give you a production path that de...
Governing What Agents Cost You
Agents break the four assumptions analytics platforms were built on. A practical guide to identity, budgets, semantic layers, caching, and instrumenta...
Every AI Model Family That Matters in Mid-2026
A full survey of the AI model landscape in mid-2026: frontier families, open-weight labs, local inference, specialists, and how to build a routing lay...
Freshness Is a Contract, Not a Note on a Dashboard
Data freshness needs to become an engineering contract with a measurable value, an owner, and consequences....
The Apache Iceberg Market in the Middle of 2026
A survey of the Apache Iceberg market in July 2026: the state of the specification, platform support, the acquisition wave, the catalog contest, and h...
The Whole Lakehouse Fits on Your Laptop Now
Consumer hardware, columnar formats, single-node engines, and the Iceberg REST catalog crossed a threshold: a large share of cluster work now runs loc...
Iceberg's Next Version Depends on Decisions Being Made in Parquet
The Iceberg and Parquet specifications are co-evolving. Several of the most consequential Iceberg v4 proposals are waiting on Parquet work....
Building a Lakehouse That Stays Inside the Border
Residency is a storage location. Sovereignty is who can compel access, who operates the systems, and whether you can leave....
Table Maintenance Stopped Being a Product
Iceberg table maintenance commoditized when every platform started shipping it. What the six operations are, what they cost, and the observability you...
The Model Class Enterprise Data Teams Are Not Using Yet
Tabular foundation models perform supervised learning in a single forward pass with no training run....
Three Vendors Are Rebuilding the Path From Transaction to Agent
Databricks, Snowflake, and SAP are closing the gap between operational databases and analytical platforms through acquisition, betting on different la...
What Zero-Copy Actually Costs
Six architectures share the phrase zero-copy, and they have different costs, failure modes, and governance stories....
The Filters We Build: Scams, Slop, and the Search for Signal, From Radio Ads to AI
How every media era from radio to AI forces humanity to build new cognitive filters, the shield against scams and the sieve for finding signal in an o...
The Breakdown of the Open Lakehouse in 2026: Iceberg, Arrow, Polaris, Parquet, and Ossie, and How to Actually Build One
A detailed breakdown of the five Apache projects powering the open lakehouse in 2026, Parquet, Iceberg, Polaris, Arrow, and Ossie, and how to build on...
Understanding the RAMpocalypse: An Economic Autopsy of the Great Memory Shortage and What Comes Next
An economic autopsy of the 2026 memory shortage, why DRAM prices surged 90% in a quarter, and what supply, demand, and market structure mean....
The Own Goal: What the Hugging Face Breach Reveals About Why Open Models Are a Security Necessity
How the 2026 Hugging Face breach proved that open models are a security necessity, closed models failed to defend, an open Chinese model succeeded....