Qiegang Long

Work notes on Lakehouse.

Variant Benchmark: Open-Sourced, Two Spark Optimizer Fixes, and a ~4x TPC-DS Speedup

We open-sourced variant-conformance-benchmark, used it to surface two Spark optimizer problems, and measured a ~4x TPC-DS speedup after the fixes — Variant finally beats string JSON on analytical queries.

July 27, 2026 · 8 min

Preliminary Notes on Open-Source Variant Performance

Early benchmark results comparing JSON-as-string vs Variant on GitHub Archive (GHA) data with Spark 4.

March 30, 2026 · 7 min