Apache Iceberg v3 Adds Native Geometry Types to Fix Decade-Old Spatial Data Gap

Apache Iceberg format version 3 introduces two native primitive types — geometry and geography — giving spatial data the same first-class treatment as timestamps and integers within lakehouse tables. For nearly a decade, teams stored coordinates either as two separate double columns or as opaque binary blobs, meaning query engines had to scan entire datasets without any spatial file-skipping. The new types pair with Apache Parquet 2.11's matching logical types to embed coordinate reference system metadata directly in the schema, eliminating silent data corruption caused by undocumented CRS assumptions. Per-file bounding-box statistics are now stored in Iceberg manifests, allowing engines to skip irrelevant files before reading any geometry data. The changes are defined at the specification level and apply across any compatible engine, not just Dremio, which currently ships Iceberg v3 support.