OnLine Analytical Processing decision support software allows the user to quickly analyze information that has been summarized into multidimensional cube views and hierarchies. OLAP tools are used to perform analysis on any type of numeric information. For example they enable users to drill down into masses of sales statistics in order to isolate products that are the most volatile.
While most OLAP tools will accept string (textual) information this is usually not very well handled and their real strength lies in numeric analysis. A classic use for OLAP is financial reporting which is by nature multidimensional - multiple months, years, accounts, regions as a minimum.
The term "on-line analytical processing" (OLAP) was coined by IBM researcher E. F. Codd in a whitepaper that set out twelve rules for analytic systems, an allusion to his earlier famous set of twelve rules defining the relational model. Quoth he:
The more sophisticated data manipulation functions such as the statistical analysis spread over multiple parts of the data base, application of complex formulae, production of reports involving data from many parts of the data base etc., were considered to be outside the realm of "basic manipulation" intended for the Relational Data Base.
By comparison with on-line transaction processing (OLTP) or relational database management systems (RDBMS), OLAP defines a technology that is optimized for processing human queries rather than transactions. The results of this orientation was that MDBMS oriented their performance requirements around a different set of benchmarks (Analytic Performance Benchmark, APB-1) than that of RDBMS (Transaction Processing Performance Council (TPC)).
As the emphasis is on Analytical Processing rather than Data Base Management some commentators do not classify OLAP systems as DBMS "databases" per se.