In computing, and more particularly in the databases field, online analytical processing (OLAP) is a type of computer-oriented local data analysis along multiple axes in order to obtain summary reports similar to those used in financial analysis. OLAP-type applications are commonly used in business intelligence, with the aim of helping management to have a transversal view of the company's activity.

This type of application is opposed to online transaction processing (OLTP) which is part of an operating system, that is, it is intended for the company's business to assist them in their management tasks.


The OLAP term was defined by Edgar Codd in 1993, who defined twelve rules that a database must respect in order to be within the OLAP concept.

- Multidimensional conceptual view
- Transparency
- Accessibility
- Consistent response times
- Client-server architecture
- Independent dimensions
- Sparse matrix management
- Multi-user access
- No restrictions on inter and intra dimensional transactions
- Easy data handling
- Simple reporting
- Unlimited number of dimensions and unlimited number of elements in dimensions

This concept was applied to a model of virtual representation of data called OLAP cube or hypercube, which can be implemented in different ways.

The OLAP hypercube gives access to information extraction functions (for visualization, analysis or processing), and to query functions in MDX language (comparable to SQL for a relational database). The results of a query are mainly read in 2 dimensions at most (array).

The following are some important concepts:

- Rotate (or Pivot): selection of the dimension pair that will form the query result
- Slice: extracting a data slice
- Scope (or Dicing): extracting a data block (a more general operation than slicing),
- Drill-up: data summary according to a dimension (example of drill-up on the time axis: going from presenting information day by day over a year, to a synthetic value for the year)
- Drill-down: it is a “zoom” equivalent; the inverse process of the drill-up
- Drill-through: when only aggregated data is available (totalized indicators), the drill-through provides access to the basic data details (particularly in the H-OLAP tools).
Variations

There are several variations similar to pilots that allow adapting data storage to different types of databases to implement the OLAP concept.

- Multidimensional OLAP (MOLAP) is an OLAP optimized for multidimensional analysis as it is based on a multidimensional warehouse. It is a form of multidimensional hypercube that allows data representation in the form of an intersection of n dimensions. These dimensions can be more or less dense, which makes the cube with more density or sparsity. Board International, IBM TM1, Essbase, Jedox Palo, Infor Alea, icCube server, Oracle OLAP, Microsoft Analysis Services and QUANTRIX are some examples of products that use MOLAP databases.

- Relational OLAP (ROLAP). In the business intelligence world, R OLAP is a data modeling and storage technique based on a relational structure. It takes advantage of existing resources (licenses, material resources, etc.) and, as such, does not require the additional investment of a multidimensional base. Examples of R-OLAP engines include Microsoft Analysis Services, MetaCube by Informix, Oracle 10g, DSS Agent by MicroStrategy and Mondrian by Pentaho.

- Hybrid OLAP (HOLAP) is a hybrid between MOLAP and ROLAP. The multidimensional structure of a hypercube is used for aggregated data. When access to a more fine-grained elementary level of detail is needed, classic relational tables are used: this is the drill through mechanism. HOLAP engine examples are Microsoft Analysis Services and Oracle OLAP.

- OLAP Space (SOLAP) is a visual platform that supports quick and easy spatiotemporal exploration and analysis of data using a multidimensional approach with multiple aggregation levels via tabular cartographic display or statistical diagram. The idea behind this is that the data representation should no longer be tabular as it happens with relational databases. With this, you should be able to present the data however you want. Laval holds a S-OLAP. There is also a site specially dedicated to SOLAP technologies: spatialbi. Several references on SOLAP are published in “The origins of the “SOLAP” term or the small story of a great idea!”. An analysis on these technologies is posted on the Intelli3 Blog - The Geodecision Software Market

- Desktop OLAP (DOLAP) is an operation mode that consists of locally retrieving part of a multidimensional database. This is interesting to continue to perform data analysis in a nomadic and disconnected way.

Originally used for instinctive data analysis, OLAP hypercubes can be coupled with data mining systems and thus analyze, predict and simulate information in a more “strictly way”. Companies generate this type of structure by massive information synchronization (ETL) from relational database management systems such as datamart, datawarehouse, or sometimes even transactional, depending on the architecture they chose for their systems.

Information is retrieved through simple requests in MDX (Multidimensional Expressions), Executive Information System (EIS), specialized applications (enterprise software) or in a spreadsheet (equipped with a navigation plugin).