OLAP — Online Analytical Processing

OLAP, or Online Analytical Processing, is a category of database technology designed for interactive analysis of large volumes of aggregated business data. It organises measures, such as revenue or quantity, against dimensions, such as time, product, region and customer, into a structure commonly visualised as a cube. This arrangement lets analysts aggregate, slice, dice and drill down through figures quickly, without writing complex queries each time. OLAP underpins many reporting and business-intelligence tools and is typically fed from a data warehouse rather than from the live transactional ERP database, so that heavy analysis does not slow day-to-day operations.

The multidimensional model

OLAP separates the numeric values being analysed, called measures or facts, from the descriptive axes used to view them, called dimensions. A dimension is usually organised into a hierarchy, for example year, quarter, month and day for time, or country, region and city for geography. The combination of measures and dimensions forms a conceptual cube. Analysts then explore it through a small set of standard operations:

• Slice — fix one dimension to a single value, such as one year.
• Dice — select a sub-cube across several dimensions.
• Drill down / roll up — move between levels of a hierarchy, from totals to detail and back.
• Pivot — rotate the cube to view different dimensions on the axes.

Because aggregates can be pre-computed, OLAP returns summarised figures far faster than running the same aggregation repeatedly against raw transactions.

OLAP versus OLTP

OLAP is the analytical counterpart to OLTP, online transaction processing, which is what an ERP system does when it records orders, postings and inventory movements. OLTP is optimised for many small, concurrent reads and writes with strict consistency; OLAP is optimised for reading and aggregating large data sets. The two have different data structures and tuning goals, which is why analytical workloads are usually offloaded from the operational system. Data is moved and reshaped from operational systems into the analytical store through an ETL process feeding a data warehouse.

Implementation variants

OLAP engines differ in how they physically store data:

• MOLAP stores data in a dedicated multidimensional structure, giving fast queries with pre-aggregation.
• ROLAP keeps data in a relational database and translates analytical queries into SQL, scaling to large data volumes.
• HOLAP combines both, holding aggregates multidimensionally and detail relationally.

In-memory column-store technology has blurred these distinctions, since some platforms perform fast aggregation directly on detailed data without a separate pre-built cube.

Relevance for SMEs

For DACH SMEs, OLAP is the analytical layer behind dashboards and management reports. It supports analyses such as period-over-period sales, margin by product line or stock turnover by warehouse, and connects naturally to tools used for OLAP analysis and to platforms such as Power BI. The main design questions are which measures and dimensions reflect the business, where the cube is fed from, and how often it is refreshed. OLAP defines the structure and engine; the act of exploring that structure is covered separately under OLAP analysis.

Related Topics

• Data warehouse
• AI in ERP
• ETL