Column-oriented Database stores data by column and optimized for Analytical (e.g. Visualization) applications which are more concern with the usage of existing data (reading, reporting, viewing, analyzing, drilling-down, exploring, interacting, visualizing) rather than with data collection (writing new data), transactional and ETL processes. In Columnar DB each column is a dictionary for one specific attribute and as such can deliver tremendous data compression over row-based data store.
Columnar DBMSes produced by multiple vendors: Sybase IQ (SAP), Vertica, KDB (Kx Systems), ParAccel, Tenbase (1010data), Infobright, etc. Column-oriented (compare with DB “sharding” ) in-memory datastores used in most advanced Data Visualization tools (e.g. Qlikview, Spotfire, Tableau), in many Data Warehouse (DW) implementations and DW appliances. Recently Columnar DB technology got even more respect, when HP bought Vertica.