Four studies on how data is laid out on disk and what compressing it actually costs. Each one takes a single question, builds whatever is needed to answer it, and settles it with measurements rather than with theory.
| Study | Question |
|---|---|
| Columnar formats against csv | What Parquet and ORC buy over plain csv, and which of the two is worth choosing |
| The SQLite file format | How a database file is actually laid out, checked by reading and writing one without the library |
| Huffman against zlib | What a hand-written entropy coder gives up to a dictionary method |
| LOUDS | What a succinct tree saves in memory and what it costs in navigation |
Every study keeps its sources, its measurements and a full report with the plots in its own directory.