Operational AI WIKI

All 0-9 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z


Online analytical processing, more commonly referred to as OLAP, is an approach to database processing. It was introduced in the 1990s as a way to use data for analytical workloads, in contrast to OLTP, which is used for transactional workloads. OLAP processing is associated with the column-oriented database, a database structure for laying out memory that is also designed for analytical workloads.


To illustrate what is meant by analytical processing on column-oriented data, imagine a company’s data in a table with rows representing customer transactions and columns representing data about each transaction (see below diagram); an OLAP query would deal with one or a few columns across very large numbers of rows. For example, summing the total amount of sales across all transactions over a given period of time would be a typical OLAP process.


OLAP entails running complex queries on large amounts of data aggregated from various sources, including OLTP databases, for the purposes of analytics and business intelligence. Year-over-year financial performance or marketing lead generation trends are examples of suitable OLAP use cases.

OLTP vs. OLAP in machine learning example

OLTP (typically row-oriented) is often contrasted with OLAP (typically column-oriented) because each approach is suited for different use cases and most databases fall into the row-oriented or column-oriented category.

graphic explaining the OLAP process in machine learning

OLAP begins with data accumulated from multiple sources and stored in a data warehouse. The data is then cleansed and stored in OLAP cubes, which users run queries against. https://searchdatamanagement.techtarget.com/definition/OLAP

OLAP at Molecula

Molecula has developed a database platform that is neither row-oriented nor column-oriented. Molecula’s feature-oriented database, FeatureBase, represents data independently of rows and columns, enabling OLAP workloads to perform faster with less hardware than the traditional columnar database. FeatureBase is particularly useful for ML and AI applications due to the nature of its size, speed, volume, and real-time data capabilities. Because of this and the fact that FeatureBase is built on a machine-native bitmap architecture, FeatureBase processing is referred to as OLMP (online machine processing).

visual diagram showing difference between row-wise vs. columnar vs. feature-oriented data format

To learn more, see FeatureBase.


Related Terms

OLAP cube

Column-oriented database


Row-oriented database

Feature-oriented database


Learn More About OLAP 

Wikipedia entry: Online analytical processing

Holistics blog: The Rise and Fall of the OLAP Cube

Tech Target: OLAP (online analytical processing) definition

Video by Hussein Nasser: Column vs Row Oriented Databases Explained