The impact of indices on the performance of data modifying queries

Sammanfattning

This thesis compares the performance of indexing over relational and semi-structured databases under updates, deletes and inserts. A total of 10 cases have been benchmarked: PostgreSQL or MongoDB databases which uses no indexing, B-tree indexing, Hash indexing, and extra two B-tree indices over non-primary attributes. The benchmark itself was done on 4 databases with 104 , 105 , 106 and 107 records generated by the Python library Faker which tries to createas unique data as possible. The results indicate that without indexing and with secondary indexing the execution time of the operations increases proportionally to the size of the database. Whilst the execution time becomes constant with B-tree and hash indexing based on primary attributes.