Bloomfilter Index

Bloomfilter Index is an equality filter acceleration index in Lakehouse. It records the presence of column values at the data file level, allowing the query engine to quickly skip files that do not contain the target value before scanning, reducing I/O.

How It Works

Bloomfilter is a probabilistic data structure that quickly determines "whether a value may exist in a certain data block":

• Not present: 100% accurate, the data block is skipped directly
• May be present: needs actual read to verify (very low probability of false positives)

Applicable Scenarios

Create Example

-- Create Bloomfilter Index at table creation CREATE TABLE orders ( order_id BIGINT, user_id BIGINT, amount DECIMAL(10,2), INDEX idx_user_id (user_id) USING BLOOMFILTER ); -- Add to an existing table ALTER TABLE orders ADD INDEX idx_order_id (order_id) USING BLOOMFILTER;

Related Documents

• Bloomfilter Index Details
• Index Usage Best Practices
• Inverted Index — full-text search scenarios