Skip to content

Häufig gestellte Fragen

What is the difference between OLAP and OLTP?
OLTP (Online Transaction Processing) refers to the operational processing of individual business transactions such as postings, orders or warehouse movements and is optimised for fast, row-by-row write access. OLAP (Online Analytical Processing), by contrast, serves downstream analytical evaluation, in which key figures are aggregated and read across large data volumes along multiple dimensions. The two worlds use differently structured data sets: the transactional ERP system on one side, an analytical layer such as a data warehouse on the other. The term OLAP was coined in 1993 by E. F. Codd, who is also credited with the relational database model.
How do MOLAP, ROLAP and HOLAP differ?
These three terms describe different technical storage architectures for OLAP. MOLAP (multidimensional OLAP) holds data in precalculated, pre-aggregated cube structures and thus delivers very fast response times, but is more memory-intensive and more rigid with very large data volumes. ROLAP (relational OLAP) leaves the data in relational tables and generates aggregates only at query time, which scales better but responds more slowly with large data volumes. HOLAP (hybrid OLAP) combines both approaches by precalculating frequently used aggregations and keeping detailed data in relational form, balancing speed and scalability.
Do I need OLAP in addition to the ERP system's standard reporting?
That depends heavily on data volume and analysis depth. As long as evaluations remain manageable, the reports integrated in the ERP are often entirely sufficient. With large historical data sets and many combined dimensions, however, direct queries on the operational tables become sluggish and put load on the production system. An analytical OLAP layer then usually sits on top of a separate data warehouse and relieves the ERP, which is why the decision should be based on the company's specific analysis needs and existing IT landscape.
Are classic OLAP cubes still state of the art?
In the classic data warehouse stack, precalculated OLAP cubes continue to be used, especially where complex in-memory calculations and very low latency are required. In modern cloud architectures they are increasingly being replaced by column-oriented (columnar) in-memory engines that perform OLAP-typical aggregations directly via SQL across millions of rows, without a dedicated cube having to be built and maintained. The decisive point is less the question of MOLAP versus columnar than the separation between transactional and analytical processing, which is preserved in both variants.
What is MDX and how does it differ from SQL?
MDX (Multidimensional Expressions) is a query language introduced by Microsoft in the late 1990s that was developed specifically for multidimensional structures such as OLAP cubes. While SQL is geared towards relational tables with rows and columns, MDX addresses dimensions, hierarchies and measures and returns multidimensional result sets. MDX has established itself as a widely used standard and is used by Microsoft Analysis Services, among others; Excel PivotTables also access cube data via MDX in the background. The language does not replace SQL but complements it for analytical questions.
Which tools and vendors are used for OLAP analyses?
The market includes analysis modules integrated in ERP and data warehouse suites as well as standalone BI and OLAP tools. Frequently mentioned examples include Microsoft Analysis Services, SAP BI solutions (such as BW/HANA), Oracle Essbase and open-source solutions such as Apache Kylin; in the self-service area, tools such as Power BI, Tableau or Looker come into play. These mentions serve as illustration and do not constitute a rating or recommendation. Which tool fits depends on the existing data architecture, the required data model and the planned maintenance effort, and should be assessed vendor-neutrally against your own requirements.