Applications of Bloom Filters

Every real use of a Bloom filter follows the same shape. Something expensive sits behind it, such as a disk read, a network call, or a full scan. The filter is asked first, and a "no" means the expensive thing can be skipped entirely.

Keep that shape in mind and the examples below stop being a list to memorize.

1. Skipping Disk Reads in a Database

This is the most common use, and the one worth knowing in detail.

Databases such as Cassandra, HBase, and the storage engines behind RocksDB and LevelDB write data into many immutable files on disk. A single key could live in any of them