B-trees balance reads and writes for OLTP. LSM trees maximize write throughput. Bitmap indexes accelerate OLAP filtering. Here is when to use each.
Iceberg's three-layer metadata tree eliminates directory listing and enables multi-level data skipping. Here is how scan planning actually works.
Databases structure data on disk as heap files, sorted files, or LSM trees, then wrap it in formats like Parquet with metadata that lets engines skip irrelevant blocks.
Iceberg uses a metadata tree, Delta Lake uses a transaction log, Hudi uses a timeline. Here is exactly how each format organizes metadata and why it matters.
