Bloom, Xor, and Cuckoo Filter Comparison for Database’s Query Optimization

This paper proposes the use of probabilistic data structure filters to optimize database systems for IT micro, small, and medium-sized enterprises (MSMEs) and startups in Indonesia. The efficiency of various filters, including the Classic Bloom filter, Partitioned Bloom filter, Counting Bloom filter, Cuckoo filter, and Xor filter, are compared in terms of their computing time, cost, and resource usage for both write and read operations. The findings show that while the use of filters led to a slight increase in insertion and query time for existing keys in the database, it also resulted in a significant decrease in query time for nonexistent keys. The Cuckoo filter was found to be the most efficient. This research provides valuable insights for IT MSMEs and startups in Indonesia, enabling them to make informed decisions in optimizing their database systems through filter selection.