Introduction to LucidDB

After years in steady-state, DBMS research and development is currently heating up again with a number of architectural innovations, foremost of which may be column-stores for read-optimized databases. LucidDB is an open source column-store DBMS that combines high performance with industrial-strength SQL processing for both queries and high-volume data integration.

This talk introduces the LucidDB project and gives an overview of its architecture. Besides column-store, other features covered include bitmap indexing, hash join/aggregation, page-level multiversioning, intelligent prefetch, star-join optimization, connectivity, and extensibility.

Benchmark results against traditional transactional DBMS projects such as MySQL will be presented, as well as recommended configurations for combining LucidDB with related projects such as Mondrian for Online Analytical Processing (OLAP).

To get the most out of the talk, attendees should have at least passing acquaintance with data warehousing concepts such as star schemas. They will be able to use what they learn from the presentation to decide whether a specialized DBMS such as LucidDB could be worth investigating as an alternative or addition to existing database implementations.