Chasing excellence, not consensus
he/him
Muhammad Ali is a Solutions Architect at ClickHouse, helping teams design high-performance analytics systems and make deliberate choices about data formats, engines, and lock-in trade-offs in the agentic era.
Apache Iceberg is one of the most important wins the data engineering community has produced — a genuine, portable foundation that gives organisations real optionality over their data. But somewhere along the way, ""avoid lock-in"" quietly became ""default to the lowest common denominator."" Most mature data organisations run multiple engines, and Iceberg absolutely earns its place as an exit strategy and interoperability contract. The mistake is treating that exit ramp as your race track. Engines like ClickHouse and Snowflake offer native formats built around fundamentally different performance characteristics — and in the agentic era, that gap is no longer academic. AI agents operate in real time, reason over fine-grained records, and are unforgiving of latency. The engineers who build the best systems won't be those who avoided every proprietary format — they'll be the ones who were intentional about where they committed and why. Use your best engine where performance is your competitive advantage. Keep Iceberg as your insurance policy. Know the difference.
