We partition the output parquet files for OSM and Overture Maps by top-level tag, which is likely the most common filtering use case.
We should also optimize the parquet files to speed up spatial queries. Some options could include one or more of the following:
- Hilbert-sorting each partitioned file
- Spatial sub-partitions
- File-level bounding boxes
- H3 or S2 metadata for each row
We partition the output parquet files for OSM and Overture Maps by top-level tag, which is likely the most common filtering use case.
We should also optimize the parquet files to speed up spatial queries. Some options could include one or more of the following: