And conversely, if storage is chunked (e.g. parquet files on S3), having time-ordered uuids may turn it into essentially an append-only log-structured store. H

And conversely, if storage is chunked (e.g. parquet files on S3), having time-ordered uuids may turn it into essentially an append-only log-structured store. Here hotspotting is the aim, since it lets you efficiently prune query plans from index scans to direct reads of the right chunk.